← Back to the catalog
Open source accessRestricted redistributionraw reproducible2 sources

OSSL AfSIS-II soil vis-NIR and MIR

ossl · NIR

OSSL AfSIS-II soil vis-NIR and MIR. Origin database composed from 2 standardized v2.0 leaves: 2 spectral source(s), 14 declared target(s). Auto-generated from dataset_card.json (verify before publication).

nirv2ossl
Original source available without a Dataverse token. The catalog uses the private tier for restricted or uncleared redistribution rights. You can obtain original resources from the source; their license still applies. Raw retrieval and automatic assembly are separate; use the loading example below.
151
samples
2,777
wavelengths
2
sources
14
targets
21
metadata
NIR
family

Provenance & citation

Where the original data and publications come from. Follow these links to obtain the source and cite its authors. Reading guide

ContributorOSSL_NIRS
Conversion statuspartial
Origin · url [open]https://storage.googleapis.com/soilspec4gg-public/ossl_soillab_L1_v1.2.csv.gz
Origin · url [open]https://storage.googleapis.com/soilspec4gg-public/ossl_visnir_L0_v1.2.csv.gz
Origin · url [open]https://storage.googleapis.com/soilspec4gg-public/ossl_soilsite_L0_v1.2.csv.gz

Governance & integrity

Source license, redistribution conditions, content version and identity hashes. Redistribution tier and original-source availability are separate. Reading guide

Tierprivate
LicenseLicenseRef-not-cleared
Permitted useResearch and benchmarking; private use only.
Access policyManual download / private-use-only per source.
RedistributionRecovered from local initial-source exports; rights not cleared for redistribution.
Content version1.1.0
Schema / protocol2.0
Content hash046780e08a0f406e…
Processing hasha3ea8ac850dc1551…
Metadata hasha75fcab3b690d7e4…

Dataset property explorer

Start with the profile axes to locate unusual variation, then inspect the measured values and evidence below. Scores are descriptive heuristics. Reading guide

Mean available profile score0.50
Highest axisIntegrity · 0.50
Diagnostics10
Sources profiled2

How to read the profile

The center is 0 and the outer ring is 1. Larger values indicate stronger anomaly or heterogeneity signals on that axis. Compare individual axes using the table; this is not an overall quality grade. Missing axes are left blank, and partial profiles have no filled polygon.

OSSL AfSIS-II soil vis-NIR and MIR property profilefewer than three profile axes available; consult the metric table

Profile axes

Dimensionless scores from 0 to 1 summarize each property; larger values indicate stronger signals. Unavailable axes are not zero.

Integrity
?Combines missing cells, zero columns and infinite values. Higher values call for checking data completeness.
0.50
Noise
?Combines roughness and its signal-to-roughness proxy. Real narrow bands may also contribute.
unavailable
Local artifacts
?Combines spike, jump and clipping heuristics. Inspect flagged spectral regions before attributing a cause.
unavailable
Baseline/shape
?Summarizes baseline slope and excess edge variation relative to signal scale.
unavailable
PCA outliers
?Combines relative PCA residual and score-distance tails. Valid rare spectra can produce a high value.
unavailable
Distance to average spectrum
?Dispersion around this dataset's mean spectrum, using distance, shape and shift summaries.
unavailable
Repeatability
?Dispersion among observations sharing an ID. Interpretation depends on the documented meaning of that ID.
unavailable
Multi-regime structure
?Variation in local density in PCA space. May reflect legitimate batches or subpopulations.
unavailable

Possible explanations to investigate

Bars rank heuristic signatures from 0 to 1. They do not estimate the probability of a physical fault. On narrow screens, scroll the chart horizontally. Full labels, scores, evidence coverage, possible causes and required external checks are in the table below.

Diagnostic hypotheses00.250.50.751hypothesis scoreSignature compatible with a spectral spliceSignature compatible with a spectral splice: 0.780.78Offset compatible with a different white referenceOffset compatible with a different white reference: 0.610.61Signature compatible with a different backgroundSignature compatible with a different background: 0.540.54Signature compatible with a population shiftSignature compatible with a population shift: 0.420.42Signature compatible with clippingSignature compatible with clipping: 0.350.35High spectral roughness signatureHigh spectral roughness signature: 0.310.31Flat spectrum/weak signalFlat spectrum/weak signal: 0.070.07

Heuristic signature scores are not calibrated probabilities of physical causes: 0.78 does not mean a 78% causal probability. Missing evidence remains unavailable. RMS/SAM describe dispersion around the dataset mean, without an independent external reference.

Diagnostic hypotheses and supporting evidence
DiagnosticScoreHeuristic strengthEvidence coverageSignalsPossible causeRequired external checks
Signature compatible with a spectral spliceX_mir0.78strong100 %Spike rate 1.00, Jump rate 1.00, Undegraded roughness S/N proxy 1.00Break at detector junctions, local calibration or different probe.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Offset compatible with a different white referenceX_mir0.61medium100 %local artifacts 1.00, RMS/SAM to mean spectrum 0.86, Baseline/mean/area 0.83Systematic discrepancy between campaigns, instruments or white reference.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with a different backgroundX_mir0.54medium100 %RMS/SAM to mean spectrum 0.86, Baseline/mean/area 0.83, Mahalanobis / T2 0.42Systematic effect of support, white/black, transflectance or measurement environment.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with a population shiftX_mir0.42medium100 %RMS/SAM to mean spectrum 0.86, PCA structure 0.58, Mahalanobis / T2 0.42Different but physically plausible variety, species, batch or condition.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Signature compatible with clippingX_mir0.35weak100 %Jump rate 1.00, Baseline/mean/area 0.83, PCA Q 0.33Detector saturated or insufficient dynamic range.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
High spectral roughness signatureX_mir0.31weak100 %Spike rate 1.00, RMS/SAM to mean spectrum 0.86, PCA Q 0.33Weak signal, detector problem, lamp or unstable acquisition.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Flat spectrum/weak signalX_mir0.07weak100 %PCA Q 0.33Bad contact, missing sample, degraded measurement or very low dynamic range.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.
Weak signatures observedX_mir—unknown90 %Weak signatures on available metrics; does not certify either physical validity or the absence of defects.Check the white/black reference actually acquired and its metadata.
Check instrument, probe, geometry, campaign and processing/resampling log.
Check the repeat protocol before interpreting an intra-ID dispersion.

Spectral sources

Each block is a separate instrument or spectral source. Read its signal summary before using PCA to explore variation between observations. Reading guide

Source coverage
SourceObservationsSample units
recovered_spectra151151
recovered_spectra151151

151 sample units are measured by every source; 0 by a single source. Sources are joined by declared sample identity, never by row position.

recovered_spectra

X_mir · NIR · unknown

Read the signal and variation summaries for this source. Axis unit: none; signal representation: not recorded.

Spectral signal summary

The mean and median describe the typical signal at each channel; shaded bands show the 25th–75th and 5th–95th percentiles across observations. A wider band means greater variation, which can be biological or instrumental.

recovered_spectra spectra1.01.52.02.53.001,0002,0003,0004,000q05-q95 envelopeq25-q75 envelopemedian spectrummedianq25–q75q05–q95wavelength / none600none — median 2.351 (q25–q75 2.235–2.445)624none — median 2.375 (q25–q75 2.247–2.508)648none — median 2.366 (q25–q75 2.222–2.492)674none — median 2.367 (q25–q75 2.224–2.513)698none — median 2.485 (q25–q75 2.354–2.626)722none — median 2.312 (q25–q75 2.171–2.45)746none — median 2.296 (q25–q75 2.14–2.445)772none — median 2.272 (q25–q75 2.144–2.417)796none — median 2.321 (q25–q75 2.183–2.479)820none — median 2.489 (q25–q75 2.385–2.588)844none — median 2.227 (q25–q75 2.121–2.351)870none — median 2.19 (q25–q75 2.062–2.304)894none — median 2.196 (q25–q75 2.066–2.318)918none — median 2.259 (q25–q75 2.123–2.406)942none — median 2.232 (q25–q75 2.074–2.392)966none — median 2.197 (q25–q75 2.055–2.333)992none — median 2.192 (q25–q75 2.073–2.304)1,016none — median 2.17 (q25–q75 2.055–2.295)1,040none — median 2.077 (q25–q75 1.968–2.189)1,064none — median 1.93 (q25–q75 1.796–2.059)1,090none — median 1.879 (q25–q75 1.713–2.053)1,114none — median 1.929 (q25–q75 1.744–2.123)1,138none — median 1.815 (q25–q75 1.677–1.979)1,162none — median 1.864 (q25–q75 1.732–2.019)1,188none — median 1.749 (q25–q75 1.633–1.91)1,212none — median 1.703 (q25–q75 1.593–1.844)1,236none — median 1.77 (q25–q75 1.64–1.921)1,260none — median 1.946 (q25–q75 1.813–2.082)1,284none — median 2.142 (q25–q75 2.047–2.244)1,310none — median 2.302 (q25–q75 2.224–2.398)1,334none — median 2.394 (q25–q75 2.276–2.521)1,358none — median 2.413 (q25–q75 2.283–2.566)1,382none — median 2.377 (q25–q75 2.266–2.536)1,408none — median 2.325 (q25–q75 2.222–2.476)1,432none — median 2.285 (q25–q75 2.174–2.395)1,456none — median 2.225 (q25–q75 2.122–2.314)1,480none — median 2.176 (q25–q75 2.088–2.272)1,506none — median 2.152 (q25–q75 2.059–2.236)1,530none — median 2.163 (q25–q75 2.068–2.233)1,554none — median 2.088 (q25–q75 2.017–2.176)1,578none — median 2.155 (q25–q75 2.076–2.239)1,602none — median 2.259 (q25–q75 2.162–2.332)1,628none — median 2.268 (q25–q75 2.186–2.364)1,652none — median 2.213 (q25–q75 2.106–2.311)1,676none — median 2.146 (q25–q75 2.078–2.253)1,700none — median 2.032 (q25–q75 1.96–2.115)1,726none — median 1.903 (q25–q75 1.835–1.963)1,750none — median 1.819 (q25–q75 1.767–1.877)1,774none — median 1.859 (q25–q75 1.811–1.918)1,798none — median 1.868 (q25–q75 1.814–1.936)1,824none — median 1.804 (q25–q75 1.745–1.866)1,848none — median 1.823 (q25–q75 1.774–1.893)1,872none — median 1.935 (q25–q75 1.835–2.022)1,896none — median 1.877 (q25–q75 1.781–1.939)1,920none — median 1.738 (q25–q75 1.687–1.817)1,946none — median 1.723 (q25–q75 1.677–1.803)1,970none — median 1.736 (q25–q75 1.669–1.804)1,994none — median 1.735 (q25–q75 1.662–1.798)2,018none — median 1.662 (q25–q75 1.616–1.726)2,044none — median 1.567 (q25–q75 1.514–1.614)2,068none — median 1.493 (q25–q75 1.441–1.549)2,092none — median 1.469 (q25–q75 1.401–1.53)2,116none — median 1.457 (q25–q75 1.385–1.518)2,142none — median 1.455 (q25–q75 1.393–1.514)2,166none — median 1.432 (q25–q75 1.353–1.493)2,190none — median 1.422 (q25–q75 1.343–1.482)2,214none — median 1.428 (q25–q75 1.371–1.485)2,238none — median 1.441 (q25–q75 1.392–1.492)2,264none — median 1.422 (q25–q75 1.366–1.476)2,288none — median 1.392 (q25–q75 1.337–1.466)2,312none — median 1.38 (q25–q75 1.315–1.456)2,336none — median 1.385 (q25–q75 1.317–1.458)2,362none — median 1.371 (q25–q75 1.302–1.457)2,386none — median 1.368 (q25–q75 1.305–1.444)2,410none — median 1.361 (q25–q75 1.295–1.442)2,434none — median 1.356 (q25–q75 1.288–1.443)2,458none — median 1.353 (q25–q75 1.286–1.444)2,484none — median 1.358 (q25–q75 1.288–1.448)2,508none — median 1.363 (q25–q75 1.291–1.451)2,532none — median 1.366 (q25–q75 1.289–1.452)2,556none — median 1.37 (q25–q75 1.288–1.455)2,582none — median 1.376 (q25–q75 1.295–1.464)2,606none — median 1.382 (q25–q75 1.3–1.472)2,630none — median 1.386 (q25–q75 1.307–1.479)2,654none — median 1.393 (q25–q75 1.316–1.49)2,680none — median 1.402 (q25–q75 1.327–1.504)2,704none — median 1.417 (q25–q75 1.337–1.517)2,728none — median 1.43 (q25–q75 1.348–1.531)2,752none — median 1.442 (q25–q75 1.36–1.546)2,776none — median 1.459 (q25–q75 1.375–1.565)2,802none — median 1.481 (q25–q75 1.392–1.588)2,826none — median 1.501 (q25–q75 1.408–1.608)2,850none — median 1.531 (q25–q75 1.433–1.64)2,874none — median 1.551 (q25–q75 1.454–1.659)2,900none — median 1.573 (q25–q75 1.477–1.685)2,924none — median 1.605 (q25–q75 1.505–1.718)2,948none — median 1.621 (q25–q75 1.519–1.737)2,972none — median 1.641 (q25–q75 1.535–1.754)2,998none — median 1.669 (q25–q75 1.555–1.778)3,022none — median 1.695 (q25–q75 1.577–1.81)3,046none — median 1.725 (q25–q75 1.6–1.847)3,070none — median 1.753 (q25–q75 1.627–1.884)3,094none — median 1.782 (q25–q75 1.653–1.916)3,120none — median 1.814 (q25–q75 1.679–1.947)3,144none — median 1.849 (q25–q75 1.705–1.974)3,168none — median 1.882 (q25–q75 1.727–2.007)3,192none — median 1.908 (q25–q75 1.745–2.036)3,218none — median 1.932 (q25–q75 1.767–2.057)3,242none — median 1.942 (q25–q75 1.778–2.069)3,266none — median 1.952 (q25–q75 1.79–2.077)3,290none — median 1.956 (q25–q75 1.799–2.084)3,316none — median 1.965 (q25–q75 1.807–2.089)3,340none — median 1.974 (q25–q75 1.816–2.1)3,364none — median 1.984 (q25–q75 1.828–2.115)3,388none — median 1.992 (q25–q75 1.835–2.123)3,412none — median 1.992 (q25–q75 1.834–2.127)3,438none — median 1.991 (q25–q75 1.838–2.131)3,462none — median 1.978 (q25–q75 1.83–2.118)3,486none — median 1.962 (q25–q75 1.822–2.102)3,510none — median 1.95 (q25–q75 1.807–2.087)3,536none — median 1.933 (q25–q75 1.793–2.075)3,560none — median 1.925 (q25–q75 1.786–2.065)3,584none — median 1.932 (q25–q75 1.782–2.088)3,608none — median 2.015 (q25–q75 1.866–2.181)3,634none — median 2.075 (q25–q75 1.917–2.261)3,658none — median 2.024 (q25–q75 1.858–2.241)3,682none — median 1.902 (q25–q75 1.737–2.137)3,706none — median 1.834 (q25–q75 1.658–2.05)3,730none — median 1.397 (q25–q75 1.316–1.518)3,756none — median 1.288 (q25–q75 1.223–1.373)3,780none — median 1.249 (q25–q75 1.196–1.325)3,804none — median 1.229 (q25–q75 1.179–1.296)3,828none — median 1.218 (q25–q75 1.17–1.283)3,854none — median 1.207 (q25–q75 1.162–1.274)3,878none — median 1.204 (q25–q75 1.158–1.267)3,902none — median 1.195 (q25–q75 1.152–1.257)3,926none — median 1.192 (q25–q75 1.149–1.251)3,952none — median 1.188 (q25–q75 1.146–1.248)3,976none — median 1.184 (q25–q75 1.143–1.245)4,000none — median 1.181 (q25–q75 1.141–1.239)

Sampling

How many observations and spectral channels were exported, and how the wavelength axis is spaced. Reading guide

Wavelengths
?Number of spectral channels. A channel can represent wavelength, wavenumber or another feature axis.
1,701
Axis range
?Lowest and highest recorded spectral-axis coordinates, in the stated unit.
600–4,000 none
Mean spacing
?Average difference between adjacent spectral-axis coordinates; compare only within the same unit.
2 none
Grid
?Uniform means adjacent axis spacings are approximately equal. An irregular grid may need care when applying derivatives.
uniform
Observations
?Exported spectral rows in this source, including repeated measurements when present.
151

Signal & quality

Signal level and roughness summaries. These depend on the signal representation and are not instrument acceptance tests. Reading guide

Value range
?Minimum and maximum recorded spectral values, in the original signal representation.
0.956 – 3.2
Mean range
?Minimum and maximum of the mean spectrum across spectral channels.
1.19 – 2.54
Mean level
?Average signal over observations and channels. The historical key mean_reflectance may also store other signal representations.
1.801
Area
?Trapezoidal integral of the mean spectrum over the sorted recorded axis; its unit is signal multiplied by axis unit.
6125
PTP
?Peak-to-peak: maximum minus minimum of the mean spectrum. Low values indicate little variation along the mean spectrum.
1.344
Noise RMS
?Historical roughness sigma proxy from robust second differences. Narrow genuine bands can also raise it; this is not repeat-measured instrument noise.
0.00013312
Roughness S/N proxy
?Average absolute signal divided by the roughness proxy. Higher means smoother signal relative to its amplitude, not certified instrument SNR.
1.4e+04
Roughness S/N proxy dB
?20 log10 of the roughness signal-to-noise proxy. Units are decibels; compare only compatible signal representations.
8e+01 dB
Dynamic range
?Maximum minus minimum of the mean spectrum. Its magnitude depends on the signal representation.
1.34
Smoothness
?Mean per-spectrum RMS of second differences, in signal units. Smaller values indicate smoother spectra under this protocol.
0.001982
Saturated
?Fraction at the observed global maximum; a heuristic indicator, not a comparison with known detector limits.
0.0%
X-outliers
?Number flagged by the qualification outlier method. An unusual spectrum can be valid and should not be discarded automatically.
63

Integrity & artefacts

Missing values, narrow spikes and abrupt jumps to investigate in the original spectra. Reading guide

NaN ratio
?Missing spectral cells divided by all spectral cells, displayed as a percentage.
0.00%
Inf count
?Number of positive or negative infinite spectral values; investigate invalid calculations or export.
0
Zero ratio
?Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing.
0.00%
Spike count
?Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged.
16,162
Spike rate
?Fraction of tested second-difference positions flagged as spikes, displayed as a percentage.
6.30%
Jump count
?Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins.
7,610
Jump rate
?Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage.
2.96%
Clip fraction
?Fraction of finite cells near the observed global minimum or maximum; inspect detector limits before concluding saturation.
0.00%

Shape & reference

Variation relative to this dataset's mean spectrum; the mean is not an independently measured calibration reference. Reading guide

Baseline slope
?Linear slope fitted to the mean spectrum over an axis normalized to 0–1. It describes total baseline change, not change per physical axis unit.
-0.74681
Curvature RMS
?Median per-spectrum RMS of second differences, summarizing curvature and high-frequency variation.
0.0019305
D1 RMS
?Median per-spectrum RMS of adjacent-channel first differences; reflects both real spectral structure and artefacts.
0.0079235
RMS to mean
?Root mean square distance to this dataset's mean spectrum, in signal units.
0.13533
RMS p95
?95th percentile of distances to the mean spectrum: 95% of observations have a distance at or below this value.
0.30065
SAM to mean
?Spectral Angle Mapper: angle to the mean spectrum, in radians. Smaller angles indicate more similar shapes.
0.052641
SAM p95
?95th percentile of spectral angles to the dataset mean, in radians. It is less sensitive to overall amplitude than RMS.
0.11228
Affine offset p95
?95th percentile of absolute fitted offsets when matching each spectrum to the dataset mean.
0.29838
Affine gain p95 Δ
?95th percentile of the absolute gain deviation from 1 in the fit to the mean spectrum.
0.15739
Affine residual p95
?95th percentile of residual RMS after gain and offset correction to the mean spectrum.
0.18608
Xcorr lag p95
?95th percentile of the absolute cross-correlation shift relative to the mean spectrum, in spectral channels.
0

Outliers & repeatability

Unusual spectra and within-ID dispersion. Repeated IDs do not by themselves establish technical repetitions. Reading guide

PCA Q p95/median
?Q (squared prediction error) is variation left outside the retained PCA subspace. The displayed ratio compares its 95th percentile with its median.
2.6
Hotelling T2 p95/median
?Hotelling T² measures distance within the retained PCA subspace, scaled by component variance. Ratio of the 95th percentile to the median.
2.8
Mahalanobis H p95/median
?Distance from the PCA-score center after variance scaling. Ratio of the 95th percentile to the median; large values identify relatively extreme spectra.
1.7
Repeat groups
?Number of IDs with several observations. Source documentation is needed to decide whether these are technical repeats, positions or time points.
0

Dimensionality (PCA)

Principal component analysis (PCA) compresses correlated wavelengths into components. This describes spectral variation, without using prediction targets. Reading guide

Effective rank
?Entropy-based summary of the variance distribution across components. Larger values indicate variation spread across more dimensions.
3.1
PCs → 95% var
?Number of principal components needed to explain 95% of spectral variance, when reached by the computed projection.
5
PCs → 99% var
?Number of principal components needed to explain 99% of spectral variance, when reached by the computed projection.
10
Top-10 cum. var
?Fraction of spectral variance explained by up to the first ten computed principal components, displayed as a percentage.
99.1%
Computed metric scores 30highest alert 1.00

Value is the measured quantity; alert score is its heuristic mapping to 0–1. Levels are weak below 0.42, medium from 0.42, and strong from 0.72. These are display bands, not calibrated acceptance thresholds. A dash means unavailable, never zero. Possible causes are suggestions to investigate. Use each metric’s ? control for its meaning and scoring: hover for a hint, or expand it with a tap, Enter or Space.

FamilyComputed metricValueAlert scoreHeuristic levelDescriptive interpretationPossible causes to verifyCalculation / scoring
Data integrityInvalid/masked cell count (owner report)
?Count of cells reported unavailable by the native owner under its null or validity mask. This count is distinct from the historical NaN ratio; consult the recorded population and denominator. Scoring: No alert mapping is defined for this metric; an unavailable score does not mean a zero measured value. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.missing_cells		——not calculableNot enough information to score this metric on this dataset.Missing data or validity mask; not evidence of an acquisition causeNative Preflight count of cells unavailable under x/null and valid_mask; denominator stays in native_metric.denominators.n_cells. Not legacy NaN ratio.
Data integrityNaN ratio
?Missing spectral cells divided by all spectral cells, displayed as a percentage. Scoring: alert = min(1, nan_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.nan_ratio		0%0.00weakFull spectrumAcquisition/export errorcount(isnan(X)) / X.sizealert = min(1, nan_ratio / 0.05)
Data integrityInf count
?Number of positive or negative infinite spectral values; investigate invalid calculations or export. Scoring: alert = min(1, inf_count / 1). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.inf_count		00.00weakNormalInvalid calculationscount(isinf(X))alert = min(1, inf_count / 1)
Data integrityZero ratio
?Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing. Scoring: alert = min(1, zero_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.zero_ratio		0%0.00weakNormalExport, saturationcount(X == 0) / count(finite X)alert = min(1, zero_ratio / 0.05)
Overall amplitudeMean reflectance
?Average signal over observations and channels. The historical key mean_reflectance may also store other signal representations. Scoring: alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.mean_reflectance		1.80140.83strongAtypical value: Too light / background visible or Too darkBackground, geometrymean(X finite)alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent
Overall amplitudeArea under curve
?Trapezoidal integral of the mean spectrum over the sorted recorded axis; its unit is signal multiplied by axis unit. Scoring: alert reuses baseline/shape drift because area scale depends on axis and units. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.area_under_curve		61250.83strongAtypical value: Illumination difference or NormalProbe distancetrapezoid(mean_spectrum, spectral_axis)alert reuses baseline/shape drift because area scale depends on axis and units
Overall amplitudePeak-to-peak (PTP)
?Peak-to-peak: maximum minus minimum of the mean spectrum. Low values indicate little variation along the mean spectrum. Scoring: alert increases when dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.peak_to_peak		1.34440.00weakStrong variabilitySaturationmax(mean_spectrum) - min(mean_spectrum)alert increases when dynamic range is abnormally flat
Overall amplitudeVariance
?Variance of all finite spectral cells, in squared signal units. It combines variation across observations and channels; a small value can describe a flat signal. Scoring: alert increases when variance/dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.variance		0.155320.00weakNormal or heterogeneousBad contactvar(X finite)alert increases when variance/dynamic range is abnormally flat
NoiseRoughness sigma proxy
?Historical roughness sigma proxy from robust second differences. Narrow genuine bands can also raise it; this is not repeat-measured instrument noise. Scoring: alert = noise_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.roughness_sigma_proxy		0.000133120.00weakStableLamp, detectormedian MAD(second derivative) * 1.4826 / sqrt(6)alert = noise_rms / signal_scale, saturated at 5%
NoiseRoughness S/N proxy
?Average absolute signal divided by the roughness proxy. Higher means smoother signal relative to its amplitude, not certified instrument SNR. Scoring: alert decreases with SNR dB; >=40 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.roughness_signal_to_noise_proxy		135320.00weakGood signalAcquisitionmean(abs(X)) / noise_rmsalert decreases with SNR dB; >=40 dB is treated as low alert
NoiseBandwise roughness S/N proxy
?Minimum channel-wise ratio of absolute mean signal to local second-difference roughness. A smaller ratio identifies a weak or rough spectral region; genuine narrow bands and near-zero signal can also lower it. This is not certified instrument SNR. Scoring: alert decreases with worst-band SNR dB; >=35 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.bandwise_roughness_signal_to_noise_proxy_min		413.420.00weakReliable zoneDetectormin(abs(mean_spectrum) / local second-derivative noise)alert decreases with worst-band SNR dB; >=35 dB is treated as low alert
Local artifactsSpike count
?Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged. Scoring: alert follows spike_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.spike_count		16,1621.00strongArtifactsCosmic rays, splicecount robust outliers in second derivativealert follows spike_rate, saturated at 1%
Local artifactsSpike rate
?Fraction of tested second-difference positions flagged as spikes, displayed as a percentage. Scoring: alert = min(1, spike_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.spike_rate		6.3%1.00strongSuspicious spectrumInterpolationspike_count / (n_samples * (n_features - 2))alert = min(1, spike_rate / 0.01)
Local artifactsJump count
?Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins. Scoring: alert follows jump_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.jump_count		7,6101.00strongDetector connectionSplicecount robust outliers in first derivativealert follows jump_rate, saturated at 1%
Local artifactsJump rate
?Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage. Scoring: alert = min(1, jump_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.jump_rate		2.96%1.00strongSpectral problemCalibrationjump_count / (n_samples * (n_features - 1))alert = min(1, jump_rate / 0.01)
Local artifactsClip fraction
?Fraction of finite cells near the observed global minimum or maximum; inspect detector limits before concluding saturation. Scoring: alert = min(1, clip_fraction / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.clip_fraction		0.000779%0.00weakNormalDetector saturatedfraction of finite cells equal to repeated min/max extremaalert = min(1, clip_fraction / 0.01)
Spectral formBaseline slope
?Linear slope fitted to the mean spectrum over an axis normalized to 0–1. It describes total baseline change, not change per physical axis unit. Scoring: alert = abs(slope / signal_scale), saturated at 0.5. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.baseline_slope		-0.746810.83strongDriftIlluminationlinear slope of mean_spectrum over normalized axisalert = abs(slope / signal_scale), saturated at 0.5
Spectral formCurvature RMS
?Median per-spectrum RMS of second differences, summarizing curvature and high-frequency variation. Scoring: alert = curvature_rms / signal_scale, saturated at 1%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.curvature_rms		0.00193050.11weakSmoothBackground, splicemedian RMS(second derivative per spectrum)alert = curvature_rms / signal_scale, saturated at 1%
Spectral formD1 RMS
?Median per-spectrum RMS of adjacent-channel first differences; reflects both real spectral structure and artefacts. Scoring: alert = d1_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.d1_rms		0.00792350.09weakFlatBiology or artifactmedian RMS(first derivative per spectrum)alert = d1_rms / signal_scale, saturated at 5%
Multivariate outliersPCA Q (SPE)
?Q (squared prediction error) is variation left outside the retained PCA subspace. The displayed ratio compares its 95th percentile with its median. Scoring: alert = min(1, pca_q_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.pca_q_ratio		2.60790.33weakCompliantArtifact, mixturep95(Q/SPE residual) / median(Q/SPE residual)alert = min(1, pca_q_ratio / 8)
Multivariate outliersHotelling T²
?Hotelling T² measures distance within the retained PCA subspace, scaled by component variance. Ratio of the 95th percentile to the median. Scoring: alert = min(1, hotelling_t2_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.hotelling_t2_ratio		2.81140.35weakCentralNatural variabilityp95(Hotelling T2) / median(Hotelling T2)alert = min(1, hotelling_t2_ratio / 8)
Multivariate outliersMahalanobis H
?Distance from the PCA-score center after variance scaling. Ratio of the 95th percentile to the median; large values identify relatively extreme spectra. Scoring: alert = min(1, mahalanobis_h_ratio / 4). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.mahalanobis_h_ratio		1.67670.42weakNormal populationDifferent domainp95(sqrt(T2)) / median(sqrt(T2))alert = min(1, mahalanobis_h_ratio / 4)
Comparison to average spectrumRMS to mean spectrum
?95th percentile of distances to the mean spectrum: 95% of observations have a distance at or below this value. Scoring: alert = RMS_p95 / signal_scale, saturated at 25%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
reference.rms_to_mean_spectrum_p95		0.300650.67mediumDifferent spectrumDomain shiftp95 RMS distance to dataset mean spectrumalert = RMS_p95 / signal_scale, saturated at 25%
Comparison to average spectrumSpectral Angle Mapper (SAM)
?95th percentile of spectral angles to the dataset mean, in radians. It is less sensitive to overall amplitude than RMS. Scoring: alert = min(1, SAM_p95 / 0.35 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
reference.sam_to_mean_spectrum_p95		0.112280.32weakSimilarBackground, geometryp95 spectral angle to dataset mean spectrumalert = min(1, SAM_p95 / 0.35 rad)
RepeatabilityRMS intra-ID
?Within-ID root mean square deviation from the group's mean spectrum, in signal units. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = RMS_intra_ID / signal_scale, saturated at 10%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.rms_intra_id		——not calculableNot enough information to score this metric on this dataset.Positioningmedian RMS distance to repeated-sample centroidalert = RMS_intra_ID / signal_scale, saturated at 10%
RepeatabilitySAM intra-ID
?Within-ID spectral angle to the group's mean spectrum, in radians. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, SAM_intra_ID / 0.15 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.sam_intra_id		——not calculableNot enough information to score this metric on this dataset.Acquisitionmedian SAM to repeated-sample centroidalert = min(1, SAM_intra_ID / 0.15 rad)
RepeatabilityCV intra-ID
?Within-ID coefficient of variation: standard deviation relative to absolute mean; unstable near zero signal. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, CV_intra_ID / 0.25). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.cv_intra_id		——not calculableNot enough information to score this metric on this dataset.Operatormedian within-ID band CValert = min(1, CV_intra_ID / 0.25)
Structure of the datasetPCA score density
?Inverse median distance to nearby observations in retained PCA-score space. Larger raw values mean more closely spaced observations; scale depends on the PCA representation. The alert uses density variation, so raw density alone does not establish an anomaly. Scoring: alert follows density_cv/profile structure complexity, not raw density alone. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.pca_score_density		0.370290.58mediumSubpopulationsDifferent batches1 / median kNN distance in PCA score spacealert follows density_cv/profile structure complexity, not raw density alone
Structure of the datasetkNN distance ratio (not LOF)
?95th percentile of mean k-nearest-neighbor distance divided by the population median of that distance, in PCA-score space. Values above 1 indicate an upper tail with more isolated observations; valid rare spectra can also be isolated. This distance ratio is not the Local Outlier Factor algorithm. Scoring: alert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.knn_distance_ratio_p95		1.7340.37weakNormal populationRare casesp95(mean kNN distance / median population mean kNN distance) in PCA-score space; not LOFalert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2)
Structure of the datasetIsolation Forest score
?95th percentile of the negated Isolation Forest score in retained PCA-score space. Larger raw scores indicate easier isolation under that model; the scale is implementation-dependent. The displayed alert reuses structure complexity rather than directly mapping this raw score. Scoring: alert follows structure complexity; raw score is implementation-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.isolation_forest_score_p95		0.548290.58mediumAtypical spectrumVarious causesp95 IsolationForest anomaly score on PCA scoresalert follows structure complexity; raw score is implementation-dependent

PCA score plot — explore spectral variation

Source recovered_spectra: 151 of 151 source observations are plotted. PC1 captures 70.2% and PC2 13.6% of the reported spectral variance: 83.8% together; 16.2% remains outside this view.

Principal component analysis (PCA) summarizes spectral differences without using targets. The horizontal PC1 axis captures the largest variation; vertical PC2 captures the next largest direction of variation. Each dot is one observation; repeated measurements need not be distinct samples. Scores are projection coordinates, not wavelengths or target values. Nearby dots have similar projections; separate groups or isolated dots are leads to inspect in the original spectra and acquisition metadata. Overlapping dots can differ in omitted dimensions. The dots have no class or batch labels, so this view cannot identify groups, confirm defective observations, explain a chemical cause or measure prediction accuracy.

recovered_spectra PCA score plot-20-1001020-10-50510PC1 -1.36 · PC2 1.819PC1 -3.941 · PC2 -1.143PC1 -9.878 · PC2 0.5175PC1 5.687 · PC2 4.554PC1 -9.565 · PC2 -1.21PC1 8.614 · PC2 4.646PC1 1.165 · PC2 1.436PC1 -4.7 · PC2 -0.4584PC1 -9.212 · PC2 0.8099PC1 4.97 · PC2 -5.526PC1 -0.0154 · PC2 -0.3827PC1 -2.703 · PC2 0.4092PC1 -3.625 · PC2 -0.6526PC1 3.525 · PC2 0.2933PC1 -4.082 · PC2 -0.9138PC1 5.937 · PC2 -2.883PC1 -12.71 · PC2 7.585PC1 0.3337 · PC2 -0.9348PC1 3.042 · PC2 1.753PC1 -0.9342 · PC2 -0.5234PC1 -5.583 · PC2 -0.06476PC1 5.522 · PC2 -2.271PC1 1.223 · PC2 -0.9313PC1 2.954 · PC2 -1.319PC1 2.17 · PC2 -0.2312PC1 -6.671 · PC2 -2.341PC1 -6.826 · PC2 -0.5579PC1 -3.289 · PC2 -3.952PC1 -3.758 · PC2 -0.6809PC1 -6.83 · PC2 -0.4133PC1 5.172 · PC2 -1.415PC1 -2.843 · PC2 -1.842PC1 8.267 · PC2 -1.988PC1 1.251 · PC2 -1.983PC1 6.416 · PC2 0.08979PC1 -1.132 · PC2 -1.736PC1 0.3464 · PC2 -2.913PC1 -0.5453 · PC2 0.8646PC1 5.611 · PC2 -0.5792PC1 4.071 · PC2 1.491PC1 3.682 · PC2 1.376PC1 9.36 · PC2 1.373PC1 -10.59 · PC2 2.734PC1 4.948 · PC2 5.075PC1 2.86 · PC2 4.763PC1 -2.197 · PC2 7.268PC1 4.754 · PC2 4.297PC1 0.7806 · PC2 -2.101PC1 1.752 · PC2 7.751PC1 -2.611 · PC2 0.08248PC1 -2.745 · PC2 -0.8341PC1 -2.942 · PC2 -1.589PC1 -6.811 · PC2 1.97PC1 0.9578 · PC2 0.08704PC1 -3.7 · PC2 -0.2186PC1 1.4 · PC2 0.6312PC1 3.626 · PC2 1.063PC1 -0.2713 · PC2 -0.8986PC1 1.863 · PC2 1.104PC1 -0.7288 · PC2 -2.962PC1 -2.779 · PC2 -0.6834PC1 -4.447 · PC2 4.724PC1 -6.461 · PC2 5.331PC1 -4.44 · PC2 4.733PC1 -11.83 · PC2 6.19PC1 -6.592 · PC2 5.993PC1 11.07 · PC2 0.8558PC1 0.1167 · PC2 1.232PC1 8.011 · PC2 3.116PC1 9.083 · PC2 2.502PC1 3.045 · PC2 -1.238PC1 5.497 · PC2 2.573PC1 1.336 · PC2 4.511PC1 16.44 · PC2 0.9299PC1 12.81 · PC2 1.35PC1 0.3303 · PC2 0.3908PC1 -10.87 · PC2 -1.038PC1 -0.1081 · PC2 -0.006473PC1 -9.428 · PC2 0.07761PC1 8.801 · PC2 2.116PC1 7.075 · PC2 0.04977PC1 11.51 · PC2 1.429PC1 -0.3784 · PC2 -0.03327PC1 -1.827 · PC2 -3.182PC1 -8.215 · PC2 -2.075PC1 -5.196 · PC2 -3.412PC1 -7.239 · PC2 4.124PC1 -6.354 · PC2 1.933PC1 3.355 · PC2 2.024PC1 -2.995 · PC2 3.138PC1 1.63 · PC2 -1.916PC1 4.362 · PC2 -0.9396PC1 -2.529 · PC2 -2.915PC1 1.314 · PC2 -1.869PC1 -8.691 · PC2 0.8917PC1 -3.906 · PC2 3.774PC1 3.872 · PC2 0.8887PC1 8.793 · PC2 3.946PC1 6.856 · PC2 3.162PC1 -2.848 · PC2 -0.9114PC1 -2.009 · PC2 -1.022PC1 -2.584 · PC2 -0.1921PC1 -4.655 · PC2 -2.951PC1 5.53 · PC2 0.6534PC1 -2.144 · PC2 -4.254PC1 -5.997 · PC2 -0.8777PC1 3.568 · PC2 -2.218PC1 2.856 · PC2 -4.851PC1 1.428 · PC2 -1.721PC1 -1.706 · PC2 -0.4027PC1 13.99 · PC2 0.4669PC1 2.374 · PC2 -0.675PC1 11.51 · PC2 -2.319PC1 1.62 · PC2 0.6733PC1 4.141 · PC2 -2.553PC1 -0.5918 · PC2 -1.15PC1 11.82 · PC2 -3.141PC1 -5.441 · PC2 -4.335PC1 -11.06 · PC2 1.148PC1 -6.135 · PC2 -1.778PC1 -9.993 · PC2 -2.034PC1 -4.297 · PC2 -2.914PC1 -2.53 · PC2 -3.249PC1 -0.4672 · PC2 -2.757PC1 -1.813 · PC2 -2.296PC1 1.809 · PC2 -2.795PC1 9.791 · PC2 0.714PC1 1.924 · PC2 -0.1497PC1 2.793 · PC2 -4.485PC1 1.158 · PC2 -2.304PC1 -6.142 · PC2 -1.884PC1 0.5017 · PC2 -5.324PC1 -7.714 · PC2 -1.492PC1 -7.367 · PC2 -1.347PC1 1.961 · PC2 0.2176PC1 0.2346 · PC2 0.172PC1 -7.668 · PC2 0.7065PC1 -0.198 · PC2 -1.544PC1 0.4948 · PC2 -3.256PC1 -7.104 · PC2 0.9009PC1 1.568 · PC2 2.111PC1 2.632 · PC2 1.247PC1 -2.357 · PC2 -0.2605PC1 -6.845 · PC2 -1.134PC1 8.435 · PC2 -1.969PC1 8.532 · PC2 -4.098PC1 -0.9097 · PC2 0.3761PC1 4.514 · PC2 0.8245PC1 -4.725 · PC2 1.333PC1 -4.781 · PC2 -2.478PC1 14.4 · PC2 1.511PC1 (70.2%)PC2 (13.6%)151 scores
View plotted PCA coordinates

Point numbers identify rows in this displayed coordinate list, not original observation or sample IDs. Non-integer coordinates use up to five significant digits. Original IDs, component loadings, preprocessing details and the observation selection method are not supplied with this plot.

Plotted observation positions (not original sample IDs)
PointPC1 scorePC2 score
1-1.36031.8188
2-3.9405-1.1432
3-9.87780.51746
45.68754.554
5-9.5654-1.2101
68.61444.6459
71.16531.4361
8-4.6998-0.45837
9-9.21250.8099
104.9705-5.5261
11-0.015402-0.38273
12-2.70250.40915
13-3.6252-0.65255
143.52550.29333
15-4.0821-0.91379
165.9374-2.8829
17-12.7137.5847
180.33366-0.93481
193.04161.7534
20-0.93425-0.52344
21-5.5832-0.064758
225.5223-2.2705
231.2226-0.93134
242.9544-1.3187
252.1701-0.23117
26-6.6713-2.3411
27-6.8262-0.55786
28-3.2892-3.9523
29-3.7579-0.68086
30-6.8296-0.41329
315.1724-1.4154
32-2.8435-1.8421
338.2671-1.988
341.2511-1.9832
356.41590.089789
36-1.1324-1.7359
370.34635-2.9133
38-0.545330.86455
395.6112-0.57919
404.07091.4914
413.68241.3764
429.35981.3731
43-10.5872.7337
444.9485.0755
452.85964.763
46-2.19687.2676
474.75444.2975
480.78063-2.1012
491.75187.751
50-2.61090.082485
51-2.7455-0.83408
52-2.9423-1.5888
53-6.81081.97
540.957760.087036
55-3.7002-0.2186
561.39970.63116
573.62571.0633
58-0.27129-0.89858
591.86271.1035
60-0.72882-2.9624
61-2.7793-0.68341
62-4.44734.7243
63-6.46155.3309
64-4.44054.7328
65-11.8276.1897
66-6.59155.9925
6711.0720.85576
680.116731.232
698.01143.1158
709.08282.5025
713.0447-1.2382
725.49722.5729
731.33574.511
7416.4420.92989
7512.8131.3496
760.330330.39079
77-10.866-1.0381
78-0.10807-0.0064731
79-9.42820.077609
808.80152.1163
817.07550.04977
8211.5081.4293
83-0.37841-0.033268
84-1.8267-3.1816
85-8.2146-2.0746
86-5.1963-3.412
87-7.23864.124
88-6.35361.9332
893.3552.0237
90-2.99453.1383
911.6299-1.9163
924.3615-0.93957
93-2.5294-2.9154
941.3136-1.8693
95-8.69140.89171
96-3.90643.7738
973.87250.88867
988.79343.9462
996.85593.1621
100-2.8482-0.91145
101-2.0093-1.0223
102-2.5836-0.19205
103-4.655-2.9513
1045.53030.65338
105-2.1443-4.2539
106-5.9967-0.87774
1073.5681-2.2177
1082.8555-4.8514
1091.4278-1.7209
110-1.7056-0.40272
11113.990.46689
1122.3739-0.67501
11311.514-2.3192
1141.61980.67333
1154.141-2.5526
116-0.59176-1.1499
11711.822-3.1406
118-5.4411-4.3354
119-11.0591.1476
120-6.1354-1.7781
121-9.9933-2.0341
122-4.2966-2.9143
123-2.5295-3.2493
124-0.46715-2.7569
125-1.8135-2.2957
1261.8087-2.7949
1279.7910.71402
1281.9239-0.14967
1292.7926-4.4852
1301.1583-2.3039
131-6.1417-1.884
1320.50168-5.324
133-7.7145-1.4924
134-7.367-1.347
1351.96120.2176
1360.234580.17203
137-7.66790.70648
138-0.19804-1.5445
1390.49482-3.2563
140-7.1040.90094
1411.56852.1109
1422.63181.2466
143-2.3575-0.26048
144-6.8448-1.1338
1458.4355-1.9692
1468.5316-4.0977
147-0.909650.37612
1484.51410.82445
149-4.72461.333
150-4.7813-2.4782
15114.3951.5107

PCA explained variance — how many dimensions?

Each bar reports the share of spectral variance explained by a component. Concentration in the first components indicates correlated wavelengths. This is not variance explained in the prediction target. This projection may use a different observation sample from the score plot.

PCA explained variance0%25%50%75%100%PC1: 70.2% (cumulative 70.2%)1PC2: 13.6% (cumulative 83.8%)2PC3: 5.7% (cumulative 89.4%)3PC4: 4.0% (cumulative 93.4%)4PC5: 2.6% (cumulative 96.0%)5PC6: 1.5% (cumulative 97.4%)6PC7: 0.6% (cumulative 98.1%)7PC8: 0.4% (cumulative 98.5%)8PC9: 0.3% (cumulative 98.8%)9PC10: 0.3% (cumulative 99.1%)10cumulative explained variancePC variancecumulativeprincipal component · cumulative (dashed)
View explained variance values
Variance per displayed principal component
ComponentSpectral variance
PC170.23%
PC213.55%
PC35.66%
PC43.97%
PC52.55%
PC61.47%
PC70.65%
PC80.37%
PC90.34%
PC100.27%
X-Y spectral correlation 14

Pearson r measures the linear association between each spectral channel (X) and a numeric target (Y), from −1 to +1. The table reports peak and average absolute correlation and the share of channels with |r| ≥ 0.5. These are in-dataset associations, not validation scores or proof of causation.

Spectral association with al_ext_usda_a1056_mg_kg

Read Pearson correlation (−1 to +1) along the spectral axis (none); the solid curve shows |r| and the dashed curve shows signed r.

X_mir · al_ext_usda_a1056_mg_kg spectral correlation-1-0.500.51absolute correlation envelopesigned correlationabsolute correlation01,0002,0003,0004,000|r|signed raxis · Pearson correlation scale

Spectral association with b_ext_mel3_mg_kg

Read Pearson correlation (−1 to +1) along the spectral axis (none); the solid curve shows |r| and the dashed curve shows signed r.

X_mir · b_ext_mel3_mg_kg spectral correlation-1-0.500.51absolute correlation envelopesigned correlationabsolute correlation01,0002,0003,0004,000|r|signed raxis · Pearson correlation scale

Spectral association with c_tot_usda_a622_w_pct

Read Pearson correlation (−1 to +1) along the spectral axis (none); the solid curve shows |r| and the dashed curve shows signed r.

X_mir · c_tot_usda_a622_w_pct spectral correlation-1-0.500.51absolute correlation envelopesigned correlationabsolute correlation01,0002,0003,0004,000|r|signed raxis · Pearson correlation scale
Targetmax |r|axis @ maxmean |r||r| ≥ .5
al_ext_usda_a1056_mg_kg0.6552,9980.46354.6%
b_ext_mel3_mg_kg0.9631,6380.57264.0%
c_tot_usda_a622_w_pct0.5382,9220.3237.1%
ca_ext_usda_a1059_mg_kg0.5883,5620.34814.5%
cu_ext_usda_a1063_mg_kg0.5143,4020.35810.9%
fe_ext_usda_a1064_mg_kg0.4031,2120.2490.0%
k_ext_usda_a1065_mg_kg0.4181,2300.2130.0%
mg_ext_usda_a1066_mg_kg0.6371,2280.3919.4%
mn_ext_usda_a1067_mg_kg0.4052,6780.2590.0%
n_tot_usda_a623_w_pct0.5432,9200.33216.6%
oc_usda_c729_w_pct0.5362,9220.3214.8%
ph_h2o_usda_a268_index0.1568140.06390.0%
s_ext_mel3_mg_kg0.3971,2080.2270.0%
zn_ext_usda_a1073_mg_kg0.2531,5540.0720.0%

recovered_spectra

X_visnir · NIR · unknown

Read the signal and variation summaries for this source. Axis unit: none; signal representation: not recorded.

Spectral signal summary

The mean and median describe the typical signal at each channel; shaded bands show the 25th–75th and 5th–95th percentiles across observations. A wider band means greater variation, which can be biological or instrumental.

recovered_spectra spectrano spectral data

Sampling

How many observations and spectral channels were exported, and how the wavelength axis is spaced. Reading guide

Wavelengths
?Number of spectral channels. A channel can represent wavelength, wavenumber or another feature axis.
1,076
Axis range
?Lowest and highest recorded spectral-axis coordinates, in the stated unit.
350–2,500 none
Mean spacing
?Average difference between adjacent spectral-axis coordinates; compare only within the same unit.
2 none
Grid
?Uniform means adjacent axis spacings are approximately equal. An irregular grid may need care when applying derivatives.
uniform
Observations
?Exported spectral rows in this source, including repeated measurements when present.
151

Signal & quality

Signal level and roughness summaries. These depend on the signal representation and are not instrument acceptance tests. Reading guide

X-outliers
?Number flagged by the qualification outlier method. An unusual spectrum can be valid and should not be discarded automatically.
0

Integrity & artefacts

Missing values, narrow spikes and abrupt jumps to investigate in the original spectra. Reading guide

NaN ratio
?Missing spectral cells divided by all spectral cells, displayed as a percentage.
100.00%
Inf count
?Number of positive or negative infinite spectral values; investigate invalid calculations or export.
0
Zero ratio
?Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing.
0.00%
Spike count
?Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged.
0
Spike rate
?Fraction of tested second-difference positions flagged as spikes, displayed as a percentage.
0.00%
Jump count
?Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins.
0
Jump rate
?Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage.
0.00%

Shape & reference

Variation relative to this dataset's mean spectrum; the mean is not an independently measured calibration reference. Reading guide

No values are available for this source in the identity card.

Outliers & repeatability

Unusual spectra and within-ID dispersion. Repeated IDs do not by themselves establish technical repetitions. Reading guide

Repeat groups
?Number of IDs with several observations. Source documentation is needed to decide whether these are technical repeats, positions or time points.
0

Dimensionality (PCA)

Principal component analysis (PCA) compresses correlated wavelengths into components. This describes spectral variation, without using prediction targets. Reading guide

No values are available for this source in the identity card.

Computed metric scores 30highest alert 1.00

Value is the measured quantity; alert score is its heuristic mapping to 0–1. Levels are weak below 0.42, medium from 0.42, and strong from 0.72. These are display bands, not calibrated acceptance thresholds. A dash means unavailable, never zero. Possible causes are suggestions to investigate. Use each metric’s ? control for its meaning and scoring: hover for a hint, or expand it with a tap, Enter or Space.

FamilyComputed metricValueAlert scoreHeuristic levelDescriptive interpretationPossible causes to verifyCalculation / scoring
Data integrityInvalid/masked cell count (owner report)
?Count of cells reported unavailable by the native owner under its null or validity mask. This count is distinct from the historical NaN ratio; consult the recorded population and denominator. Scoring: No alert mapping is defined for this metric; an unavailable score does not mean a zero measured value. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.missing_cells		——not calculableNot enough information to score this metric on this dataset.Missing data or validity mask; not evidence of an acquisition causeNative Preflight count of cells unavailable under x/null and valid_mask; denominator stays in native_metric.denominators.n_cells. Not legacy NaN ratio.
Data integrityNaN ratio
?Missing spectral cells divided by all spectral cells, displayed as a percentage. Scoring: alert = min(1, nan_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.nan_ratio		100%1.00strongCorrupted spectrumAcquisition/export errorcount(isnan(X)) / X.sizealert = min(1, nan_ratio / 0.05)
Data integrityInf count
?Number of positive or negative infinite spectral values; investigate invalid calculations or export. Scoring: alert = min(1, inf_count / 1). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.inf_count		00.00weakNormalInvalid calculationscount(isinf(X))alert = min(1, inf_count / 1)
Data integrityZero ratio
?Exactly zero spectral cells divided by all cells. Zero can be a valid measurement and is not automatically missing. Scoring: alert = min(1, zero_ratio / 0.05). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
integrity.zero_ratio		0%0.00weakNormalExport, saturationcount(X == 0) / count(finite X)alert = min(1, zero_ratio / 0.05)
Overall amplitudeMean reflectance
?Average signal over observations and channels. The historical key mean_reflectance may also store other signal representations. Scoring: alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.mean_reflectance		——not calculableNot enough information to score this metric on this dataset.Background, geometrymean(X finite)alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent
Overall amplitudeArea under curve
?Trapezoidal integral of the mean spectrum over the sorted recorded axis; its unit is signal multiplied by axis unit. Scoring: alert reuses baseline/shape drift because area scale depends on axis and units. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.area_under_curve		——not calculableNot enough information to score this metric on this dataset.Probe distancetrapezoid(mean_spectrum, spectral_axis)alert reuses baseline/shape drift because area scale depends on axis and units
Overall amplitudePeak-to-peak (PTP)
?Peak-to-peak: maximum minus minimum of the mean spectrum. Low values indicate little variation along the mean spectrum. Scoring: alert increases when dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.peak_to_peak		——not calculableNot enough information to score this metric on this dataset.Saturationmax(mean_spectrum) - min(mean_spectrum)alert increases when dynamic range is abnormally flat
Overall amplitudeVariance
?Variance of all finite spectral cells, in squared signal units. It combines variation across observations and channels; a small value can describe a flat signal. Scoring: alert increases when variance/dynamic range is abnormally flat. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
amplitude.variance		——not calculableNot enough information to score this metric on this dataset.Bad contactvar(X finite)alert increases when variance/dynamic range is abnormally flat
NoiseRoughness sigma proxy
?Historical roughness sigma proxy from robust second differences. Narrow genuine bands can also raise it; this is not repeat-measured instrument noise. Scoring: alert = noise_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.roughness_sigma_proxy		——not calculableNot enough information to score this metric on this dataset.Lamp, detectormedian MAD(second derivative) * 1.4826 / sqrt(6)alert = noise_rms / signal_scale, saturated at 5%
NoiseRoughness S/N proxy
?Average absolute signal divided by the roughness proxy. Higher means smoother signal relative to its amplitude, not certified instrument SNR. Scoring: alert decreases with SNR dB; >=40 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.roughness_signal_to_noise_proxy		——not calculableNot enough information to score this metric on this dataset.Acquisitionmean(abs(X)) / noise_rmsalert decreases with SNR dB; >=40 dB is treated as low alert
NoiseBandwise roughness S/N proxy
?Minimum channel-wise ratio of absolute mean signal to local second-difference roughness. A smaller ratio identifies a weak or rough spectral region; genuine narrow bands and near-zero signal can also lower it. This is not certified instrument SNR. Scoring: alert decreases with worst-band SNR dB; >=35 dB is treated as low alert. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
noise.bandwise_roughness_signal_to_noise_proxy_min		——not calculableNot enough information to score this metric on this dataset.Detectormin(abs(mean_spectrum) / local second-derivative noise)alert decreases with worst-band SNR dB; >=35 dB is treated as low alert
Local artifactsSpike count
?Local narrow peaks flagged by the second-difference heuristic; real sharp spectral bands may also be flagged. Scoring: alert follows spike_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.spike_count		00.00weakClean spectrumCosmic rays, splicecount robust outliers in second derivativealert follows spike_rate, saturated at 1%
Local artifactsSpike rate
?Fraction of tested second-difference positions flagged as spikes, displayed as a percentage. Scoring: alert = min(1, spike_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.spike_rate		0%0.00weakNormalInterpolationspike_count / (n_samples * (n_features - 2))alert = min(1, spike_rate / 0.01)
Local artifactsJump count
?Adjacent-channel changes flagged by the first-difference heuristic, potentially including detector joins. Scoring: alert follows jump_rate, saturated at 1%. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.jump_count		00.00weakContinuousSplicecount robust outliers in first derivativealert follows jump_rate, saturated at 1%
Local artifactsJump rate
?Fraction of tested adjacent-channel differences flagged as jumps, displayed as a percentage. Scoring: alert = min(1, jump_rate / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.jump_rate		0%0.00weakNormalCalibrationjump_count / (n_samples * (n_features - 1))alert = min(1, jump_rate / 0.01)
Local artifactsClip fraction
?Fraction of finite cells near the observed global minimum or maximum; inspect detector limits before concluding saturation. Scoring: alert = min(1, clip_fraction / 0.01). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
artefacts.clip_fraction		——not calculableNot enough information to score this metric on this dataset.Detector saturatedfraction of finite cells equal to repeated min/max extremaalert = min(1, clip_fraction / 0.01)
Spectral formBaseline slope
?Linear slope fitted to the mean spectrum over an axis normalized to 0–1. It describes total baseline change, not change per physical axis unit. Scoring: alert = abs(slope / signal_scale), saturated at 0.5. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.baseline_slope		——not calculableNot enough information to score this metric on this dataset.Illuminationlinear slope of mean_spectrum over normalized axisalert = abs(slope / signal_scale), saturated at 0.5
Spectral formCurvature RMS
?Median per-spectrum RMS of second differences, summarizing curvature and high-frequency variation. Scoring: alert = curvature_rms / signal_scale, saturated at 1%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.curvature_rms		——not calculableNot enough information to score this metric on this dataset.Background, splicemedian RMS(second derivative per spectrum)alert = curvature_rms / signal_scale, saturated at 1%
Spectral formD1 RMS
?Median per-spectrum RMS of adjacent-channel first differences; reflects both real spectral structure and artefacts. Scoring: alert = d1_rms / signal_scale, saturated at 5%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
shape.d1_rms		——not calculableNot enough information to score this metric on this dataset.Biology or artifactmedian RMS(first derivative per spectrum)alert = d1_rms / signal_scale, saturated at 5%
Multivariate outliersPCA Q (SPE)
?Q (squared prediction error) is variation left outside the retained PCA subspace. The displayed ratio compares its 95th percentile with its median. Scoring: alert = min(1, pca_q_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.pca_q_ratio		——not calculableNot enough information to score this metric on this dataset.Artifact, mixturep95(Q/SPE residual) / median(Q/SPE residual)alert = min(1, pca_q_ratio / 8)
Multivariate outliersHotelling T²
?Hotelling T² measures distance within the retained PCA subspace, scaled by component variance. Ratio of the 95th percentile to the median. Scoring: alert = min(1, hotelling_t2_ratio / 8). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.hotelling_t2_ratio		——not calculableNot enough information to score this metric on this dataset.Natural variabilityp95(Hotelling T2) / median(Hotelling T2)alert = min(1, hotelling_t2_ratio / 8)
Multivariate outliersMahalanobis H
?Distance from the PCA-score center after variance scaling. Ratio of the 95th percentile to the median; large values identify relatively extreme spectra. Scoring: alert = min(1, mahalanobis_h_ratio / 4). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
outliers.mahalanobis_h_ratio		——not calculableNot enough information to score this metric on this dataset.Different domainp95(sqrt(T2)) / median(sqrt(T2))alert = min(1, mahalanobis_h_ratio / 4)
Comparison to average spectrumRMS to mean spectrum
?95th percentile of distances to the mean spectrum: 95% of observations have a distance at or below this value. Scoring: alert = RMS_p95 / signal_scale, saturated at 25%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
reference.rms_to_mean_spectrum_p95		——not calculableNot enough information to score this metric on this dataset.Domain shiftp95 RMS distance to dataset mean spectrumalert = RMS_p95 / signal_scale, saturated at 25%
Comparison to average spectrumSpectral Angle Mapper (SAM)
?95th percentile of spectral angles to the dataset mean, in radians. It is less sensitive to overall amplitude than RMS. Scoring: alert = min(1, SAM_p95 / 0.35 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
reference.sam_to_mean_spectrum_p95		——not calculableNot enough information to score this metric on this dataset.Background, geometryp95 spectral angle to dataset mean spectrumalert = min(1, SAM_p95 / 0.35 rad)
RepeatabilityRMS intra-ID
?Within-ID root mean square deviation from the group's mean spectrum, in signal units. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = RMS_intra_ID / signal_scale, saturated at 10%. Signal scale is the larger of the absolute mean level and absolute peak-to-peak range, with a floor of 1e-12. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.rms_intra_id		——not calculableNot enough information to score this metric on this dataset.Positioningmedian RMS distance to repeated-sample centroidalert = RMS_intra_ID / signal_scale, saturated at 10%
RepeatabilitySAM intra-ID
?Within-ID spectral angle to the group's mean spectrum, in radians. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, SAM_intra_ID / 0.15 rad). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.sam_intra_id		——not calculableNot enough information to score this metric on this dataset.Acquisitionmedian SAM to repeated-sample centroidalert = min(1, SAM_intra_ID / 0.15 rad)
RepeatabilityCV intra-ID
?Within-ID coefficient of variation: standard deviation relative to absolute mean; unstable near zero signal. Repeated IDs require documentation before this spread can be called technical repeatability. Scoring: alert = min(1, CV_intra_ID / 0.25). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
repeatability.cv_intra_id		——not calculableNot enough information to score this metric on this dataset.Operatormedian within-ID band CValert = min(1, CV_intra_ID / 0.25)
Structure of the datasetPCA score density
?Inverse median distance to nearby observations in retained PCA-score space. Larger raw values mean more closely spaced observations; scale depends on the PCA representation. The alert uses density variation, so raw density alone does not establish an anomaly. Scoring: alert follows density_cv/profile structure complexity, not raw density alone. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.pca_score_density		——not calculableNot enough information to score this metric on this dataset.Different batches1 / median kNN distance in PCA score spacealert follows density_cv/profile structure complexity, not raw density alone
Structure of the datasetkNN distance ratio (not LOF)
?95th percentile of mean k-nearest-neighbor distance divided by the population median of that distance, in PCA-score space. Values above 1 indicate an upper tail with more isolated observations; valid rare spectra can also be isolated. This distance ratio is not the Local Outlier Factor algorithm. Scoring: alert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2). Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.knn_distance_ratio_p95		——not calculableNot enough information to score this metric on this dataset.Rare casesp95(mean kNN distance / median population mean kNN distance) in PCA-score space; not LOFalert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2)
Structure of the datasetIsolation Forest score
?95th percentile of the negated Isolation Forest score in retained PCA-score space. Larger raw scores indicate easier isolation under that model; the scale is implementation-dependent. The displayed alert reuses structure complexity rather than directly mapping this raw score. Scoring: alert follows structure complexity; raw score is implementation-dependent. Higher alert scores indicate stronger heuristic concern, not a fault probability or an acceptance decision. A dash means unavailable.
structure.isolation_forest_score_p95		——not calculableNot enough information to score this metric on this dataset.Various causesp95 IsolationForest anomaly score on PCA scoresalert follows structure complexity; raw score is implementation-dependent

PCA score plot — explore spectral variation

PC1 and PC2 summarize differences between observations without using targets; their coordinates are not wavelengths or target values.

A usable two-component observation projection is unavailable in this view.

PCA explained variance — how many dimensions?

Each bar reports the share of spectral variance explained by a component. Concentration in the first components indicates correlated wavelengths. This is not variance explained in the prediction target. This projection may use a different observation sample from the score plot.

PCA explained variancePCA unavailable
View explained variance values

Metric interpretation reference

Definitions, formulas and possible explanations for the reported values. A possible cause requires verification against acquisition records. Reading guide

Metric catalog 30

Look up each measured quantity, its calculation and the heuristic alert mapping before interpreting a high or low value.

FamilyMetricWhat it measuresHigher values suggestLower values suggestPossible causesCalculation / score
Data integrityInvalid/masked cell count (owner report)Null or masked cells in the evaluated populationTo investigateComplete according to the declared maskMissing data or validity mask; not evidence of an acquisition causeNative Preflight count of cells unavailable under x/null and valid_mask; denominator stays in native_metric.denominators.n_cells. Not legacy NaN ratio.
Data integrityNaN ratioMissing dataCorrupted spectrumFull spectrumAcquisition/export errorcount(isnan(X)) / X.sizealert = min(1, nan_ratio / 0.05)
Data integrityInf countInfinite valuesCorruptionNormalInvalid calculationscount(isinf(X))alert = min(1, inf_count / 1)
Data integrityZero ratioNull columns or cellsTruncated spectrumNormalExport, saturationcount(X == 0) / count(finite X)alert = min(1, zero_ratio / 0.05)
Overall amplitudeMean reflectanceAverage levelToo light / background visibleToo darkBackground, geometrymean(X finite)alert reuses baseline/shape drift because absolute reflectance ranges are technology-dependent
Overall amplitudeArea under curveOverall intensityIllumination differenceNormalProbe distancetrapezoid(mean_spectrum, spectral_axis)alert reuses baseline/shape drift because area scale depends on axis and units
Overall amplitudePeak-to-peak (PTP)Dynamic rangeStrong variabilityFlat spectrumSaturationmax(mean_spectrum) - min(mean_spectrum)alert increases when dynamic range is abnormally flat
Overall amplitudeVarianceSpectral variabilityNormal or heterogeneousFlat spectrumBad contactvar(X finite)alert increases when variance/dynamic range is abnormally flat
NoiseRoughness sigma proxyHigh frequency noiseNoisyStableLamp, detectormedian MAD(second derivative) * 1.4826 / sqrt(6)alert = noise_rms / signal_scale, saturated at 5%
NoiseRoughness S/N proxySignal qualityGood signalBad signalAcquisitionmean(abs(X)) / noise_rmsalert decreases with SNR dB; >=40 dB is treated as low alert
NoiseBandwise roughness S/N proxyLocalized noiseReliable zoneProblem areaDetectormin(abs(mean_spectrum) / local second-derivative noise)alert decreases with worst-band SNR dB; >=35 dB is treated as low alert
Local artifactsSpike countNarrow peaksArtifactsClean spectrumCosmic rays, splicecount robust outliers in second derivativealert follows spike_rate, saturated at 1%
Local artifactsSpike ratePeak densitySuspicious spectrumNormalInterpolationspike_count / (n_samples * (n_features - 2))alert = min(1, spike_rate / 0.01)
Local artifactsJump countDiscontinuitiesDetector connectionContinuousSplicecount robust outliers in first derivativealert follows jump_rate, saturated at 1%
Local artifactsJump rateJump frequencySpectral problemNormalCalibrationjump_count / (n_samples * (n_features - 1))alert = min(1, jump_rate / 0.01)
Local artifactsClip fractionSaturationClippingNormalDetector saturatedfraction of finite cells equal to repeated min/max extremaalert = min(1, clip_fraction / 0.01)
Spectral formBaseline slopeOverall slopeDriftStableIlluminationlinear slope of mean_spectrum over normalized axisalert = abs(slope / signal_scale), saturated at 0.5
Spectral formCurvature RMSCurvatureUnusual shapeSmoothBackground, splicemedian RMS(second derivative per spectrum)alert = curvature_rms / signal_scale, saturated at 1%
Spectral formD1 RMSLocal variabilityStructured spectrumFlatBiology or artifactmedian RMS(first derivative per spectrum)alert = d1_rms / signal_scale, saturated at 5%
Multivariate outliersPCA Q (SPE)Not explained by PCAAtypical spectrumCompliantArtifact, mixturep95(Q/SPE residual) / median(Q/SPE residual)alert = min(1, pca_q_ratio / 8)
Multivariate outliersHotelling T²Extreme in PCAExtreme but consistentCentralNatural variabilityp95(Hotelling T2) / median(Hotelling T2)alert = min(1, hotelling_t2_ratio / 8)
Multivariate outliersMahalanobis HDistance to cloudGlobal outlierNormal populationDifferent domainp95(sqrt(T2)) / median(sqrt(T2))alert = min(1, mahalanobis_h_ratio / 4)
Comparison to average spectrumRMS to mean spectrumAverage distanceDifferent spectrumTypicalDomain shiftp95 RMS distance to dataset mean spectrumalert = RMS_p95 / signal_scale, saturated at 25%
Comparison to average spectrumSpectral Angle Mapper (SAM)Difference in shapeDifferent shapeSimilarBackground, geometryp95 spectral angle to dataset mean spectrumalert = min(1, SAM_p95 / 0.35 rad)
RepeatabilityRMS intra-IDReproducibilityPoor repeatabilityStablePositioningmedian RMS distance to repeated-sample centroidalert = RMS_intra_ID / signal_scale, saturated at 10%
RepeatabilitySAM intra-IDShape variationUnstableStableAcquisitionmedian SAM to repeated-sample centroidalert = min(1, SAM_intra_ID / 0.15 rad)
RepeatabilityCV intra-IDInternal variabilityBad controlStableOperatormedian within-ID band CValert = min(1, CV_intra_ID / 0.25)
Structure of the datasetPCA score densityClustersSubpopulationsHomogeneousDifferent batches1 / median kNN distance in PCA score spacealert follows density_cv/profile structure complexity, not raw density alone
Structure of the datasetkNN distance ratio (not LOF)Local anomalyIsolated spectrumNormal populationRare casesp95(mean kNN distance / median population mean kNN distance) in PCA-score space; not LOFalert = min(1, max(0, kNN_distance_ratio_p95 - 1) / 2)
Structure of the datasetIsolation Forest scoreGlobal anomalyAtypical spectrumNormalVarious causesp95 IsolationForest anomaly score on PCA scoresalert follows structure complexity; raw score is implementation-dependent
Technology-specific extensions

Find measurement-specific checks and acquisition context that can help interpret a source's spectral patterns.

TechnologyAdaptations / metricsTarget anomaliesPractical guidance
UV-Vis 300-1000 nmBaseline, overall slope, drift at edges 300-350 and 900-1000; metrics by zonesStray light, bad white, saturation, weak signal at the endsThe edges are often unstable; also calculate edge/middle scores.
UV-Vis 300-1000 nmSaturation / clipping near max absorbance or max reflectanceSignal clippedVery important if strong absorption.
UV-Vis 300-1000 nmRed-edge, peak position, band ratios for vegetationBiological shift or optical artifactHelps distinguish real change from acquisition problem.
UV-Vis 300-1000 nmSmoothness / roughness indexHigh frequency noiseOften more informative than the SNR alone.
MIR / ATR-FTIRATR contact quality index: overall intensity, total area, depth of key bandsPoor crystal-sample contactCrucial: many anomalies come from the ATR contact.
MIR / ATR-FTIRCO2 / H2O atmospheric bandsBad atmospheric correctionFrequent spurious peaks.
MIR / ATR-FTIRBaseline curvature / rubber-band residualDiffusion, contact, baseline driftVery useful before PCA.
MIR / ATR-FTIRPeak position shiftBad spectral alignment/calibrationImportant in FTIR because small shifts count.
MIR / ATR-FTIRBand area ratios on known bandsChemically incoherent spectrumTo be adapted by matrix: polysaccharides, proteins, lipids, etc.
HS-MSTotal Ion Current (TIC), Base Peak Intensity (BPI)Weak injection, unstable ionizationMS equivalent of the global spectral level.
HS-MSNumber of peaks detectedPoor or too noisy spectrumToo little = bad signal; too much = noise/contamination.
HS-MSMass accuracy / m/z driftMass calibration problemFundamental in HRMS.
HS-MSRetention time drift if LC/GC-MSChromatographic driftMonitor on standards/QC pools.
HS-MSBlank contamination scoreContaminants / carry-overCompare samples vs blanks.
HS-MSInternal standard CVInstrumental variabilityVery robust if standards available.
HS-MSMissingness by featureDetection instabilityCrucial for filtering variables.
With repetitionsIntra-sample RMSOverall repeatabilityApplicable to all technologies.
With repetitionsSAM / intra-sample correlationShape repeatabilityVery useful for spectra.
With repetitionsIntra-sample CV per band / featureLocal repeatabilityDetects unstable areas.
With repetitionsICC or variance componentsShare of sample vs. technical varianceVery useful if several repetitions per sample.
With repetitionsDistance to intra-ID centroidAberrant repetitionAllows you to flag the bad repetition rather than the entire sample.
Bug-hunting / supervised audits

Review suggested follow-up methods for suspected data issues and whether their results are available in this explorer.

Potential issue familySuggested methodsWhat it detectsExplorer availability
Global spectral shiftInter-dataset average spectrum correlation, DTW, cross-correlation, comparison of peak positionsWavelength shift, misalignment, different interpolationPartially calculated: cross-correlation lag and dispersion of peak positions vs. average spectrum.
Baseline / offset / gainRegression each spectrum vs average spectrum: x = a + b ref + residual; followed by a, b, residual RMSAdditive offset, multiplicative effect, baseline driftCalculated in reference.affine_*.
Row mixing / X-M-Y mismatchIndex check, row hash, ID duplication, intra-ID spectral distance, inconsistent labelsMixed up rows, misaligned metadata, Y assigned to wrong spectrumPartially covered by intra-ID repeatability; checks index/hash to add to the canonical pipeline.
Information leak / poorly split repetitionsGroupKFold by sample_id vs StratifiedKFold random; auditing partitions by sample_idArtificially good performance due to repetitionsRequires splits and benchmark model; not calculated by the descriptive card.
Label bugsClose samples in X but different Y, confident learning, systematic FP/FN errorsInverted Y, input errors, ambiguous classesRequires Y and/or model; recommended for supervised explorer.
Hidden subdomainsPCA/UMAP/t-SNE + unsupervised clustering + association with dataset/Y/date/operatorBatches, campaigns, probes, backgrounds not providedPartially calculated by PCA and distance ratio kNN (not LOF); UMAP/t-SNE outside the static card.
Unknown localized artifactsWavelength x dataset map: mean difference, variance difference, KS per wavelengthUnanticipated anomalous spectral regionsTo be calculated at the bank level when several datasets share a spectral axis.
Instrumental breaksDiscontinuities in derivatives, changepoint detectionSplice, detector connection, local jump not providedCalculated by jump/spike rates; more advanced changepoint to add.
Spectral mixing/contaminationNMF / unmixing / reconstruction by convex hullExternal component: background, plastic, soilNot calculated automatically; requires component assumptions or large library.
Unstable but predictive featuresModel importance vs. QC instability by variableModel that learns an artifact rather than a biological signalRequires supervised model; recommended for benchmark reporting.

Curated scientific variables

Scientific roles derived from the source contract. M denotes experimental metadata; Y denotes informative response variables. Reading guide

0 experimental metadata fields (M) · 14 informative targets (Y).

VariableRoleDistinct valuesCompleteRange / balance
al_ext_usda_a1056_mg_kgY145100%108 – 1,672
b_ext_mel3_mg_kgY44.636%0.3 – 1
c_tot_usda_a622_w_pctY142100%0.097 – 4.406
ca_ext_usda_a1059_mg_kgY139100%15 – 4,833
cu_ext_usda_a1063_mg_kgY3899.34%0.1 – 6.29
fe_ext_usda_a1064_mg_kgY113100%24 – 578
k_ext_usda_a1065_mg_kgY8799.34%6 – 416
mg_ext_usda_a1066_mg_kgY121100%2 – 1,351
mn_ext_usda_a1067_mg_kgY144100%0.7 – 509.9
n_tot_usda_a623_w_pctY94100%0.008 – 0.428
oc_usda_c729_w_pctY142100%0.097 – 4.396
ph_h2o_usda_a268_indexY109100%4.1 – 8.32
s_ext_mel3_mg_kgY25100%1 – 84
zn_ext_usda_a1073_mg_kgY4191.39%0.09 – 30.8

Identifiers and sampling groups remain available separately. Statistics count exported observations.

Variables

Declared prediction targets and recorded metadata, with distributions and missing-value counts. Identifiers are not automatically prediction targets. Reading guide

Targets 14

al_ext_usda_a1056_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

al_ext_usda_a1056_mg_kg distribution02040108 – 173.2: 6173.2 – 238.3: 10238.3 – 303.5: 16303.5 – 368.7: 14368.7 – 433.8: 18433.8 – 499: 28499 – 564.2: 28564.2 – 629.3: 32629.3 – 694.5: 20694.5 – 759.7: 16759.7 – 824.8: 16824.8 – 890: 22890 – 955.2: 14955.2 – 1020: 121020 – 1086: 161086 – 1151: 81151 – 1216: 41216 – 1,281: 21,281 – 1346: 41346 – 1411: 61411 – 1476: 01476 – 1542: 21542 – 1607: 21607 – 1,672: 605001,0001,5002,000
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
694.1 ± 333
Median
?50th percentile: half the non-missing values lie at or below it.
628
Range
?Smallest and largest recorded non-missing values.
108 – 1,672
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.479
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
0.78 / 0.43
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

b_ext_mel3_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

b_ext_mel3_mg_kg distribution02460.3 – 0.3292: 20.3292 – 0.3583: 00.3583 – 0.3875: 00.3875 – 0.4167: 00.4167 – 0.4458: 00.4458 – 0.475: 00.475 – 0.5042: 20.5042 – 0.5333: 00.5333 – 0.5625: 00.5625 – 0.5917: 00.5917 – 0.6208: 00.6208 – 0.65: 00.65 – 0.6792: 00.6792 – 0.7083: 00.7083 – 0.7375: 00.7375 – 0.7667: 00.7667 – 0.7958: 00.7958 – 0.825: 40.825 – 0.8542: 00.8542 – 0.8833: 00.8833 – 0.9125: 00.9125 – 0.9417: 00.9417 – 0.9708: 00.9708 – 1: 60.10.20.51
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 288
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
0.7714 ± 0.264
Median
?50th percentile: half the non-missing values lie at or below it.
0.8
Range
?Smallest and largest recorded non-missing values.
0.3 – 1
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.343
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
-0.86 / -0.68
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
yes

c_tot_usda_a622_w_pct

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

c_tot_usda_a622_w_pct distribution02550750.097 – 0.2765: 540.2765 – 0.4561: 620.4561 – 0.6356: 440.6356 – 0.8152: 300.8152 – 0.9947: 220.9947 – 1.174: 241.174 – 1.354: 241.354 – 1.533: 161.533 – 1.713: 61.713 – 1.892: 21.892 – 2.072: 42.072 – 2.251: 22.251 – 2.431: 02.431 – 2.611: 02.611 – 2.79: 22.79 – 2.97: 22.97 – 3.149: 03.149 – 3.329: 23.329 – 3.508: 23.508 – 3.688: 03.688 – 3.867: 03.867 – 4.047: 24.047 – 4.226: 04.226 – 4.406: 20246
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
0.8046 ± 0.712
Median
?50th percentile: half the non-missing values lie at or below it.
0.587
Range
?Smallest and largest recorded non-missing values.
0.097 – 4.406
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.885
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
2.4 / 7.6
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

ca_ext_usda_a1059_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

ca_ext_usda_a1059_mg_kg distribution05010015 – 215.8: 80215.8 – 416.5: 48416.5 – 617.2: 38617.2 – 818: 22818 – 1019: 281019 – 1220: 161220 – 1420: 161420 – 1,621: 81,621 – 1822: 61822 – 2022: 102022 – 2223: 42223 – 2,424: 22,424 – 2625: 62625 – 2826: 42826 – 3026: 43026 – 3,227: 23,227 – 3428: 23428 – 3628: 03628 – 3829: 03829 – 4,030: 24,030 – 4231: 04231 – 4432: 24432 – 4632: 04632 – 4,833: 202,0004,0006,000
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
833.8 ± 895
Median
?50th percentile: half the non-missing values lie at or below it.
525
Range
?Smallest and largest recorded non-missing values.
15 – 4,833
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
1.07
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
1.9 / 3.9
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

cu_ext_usda_a1063_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

cu_ext_usda_a1063_mg_kg distribution0501000.1 – 0.3579: 780.3579 – 0.6158: 620.6158 – 0.8738: 260.8738 – 1.132: 321.132 – 1.39: 181.39 – 1.648: 221.648 – 1.905: 121.905 – 2.163: 82.163 – 2.421: 62.421 – 2.679: 42.679 – 2.937: 42.937 – 3.195: 63.195 – 3.453: 63.453 – 3.711: 03.711 – 3.969: 03.969 – 4.227: 84.227 – 4.485: 24.485 – 4.742: 04.742 – 5: 25 – 5.258: 05.258 – 5.516: 25.516 – 5.774: 05.774 – 6.032: 06.032 – 6.29: 202468
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 2
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
1.128 ± 1.16
Median
?50th percentile: half the non-missing values lie at or below it.
0.7
Range
?Smallest and largest recorded non-missing values.
0.1 – 6.29
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
1.03
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
1.9 / 4
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

fe_ext_usda_a1064_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

fe_ext_usda_a1064_mg_kg distribution025507524 – 47.08: 3647.08 – 70.17: 4670.17 – 93.25: 6293.25 – 116.3: 44116.3 – 139.4: 16139.4 – 162.5: 16162.5 – 185.6: 14185.6 – 208.7: 12208.7 – 231.8: 8231.8 – 254.8: 10254.8 – 277.9: 8277.9 – 301: 2301 – 324.1: 10324.1 – 347.2: 0347.2 – 370.2: 2370.2 – 393.3: 2393.3 – 416.4: 4416.4 – 439.5: 2439.5 – 462.6: 2462.6 – 485.7: 0485.7 – 508.8: 2508.8 – 531.8: 0531.8 – 554.9: 0554.9 – 578: 40200400600
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
135.9 ± 108
Median
?50th percentile: half the non-missing values lie at or below it.
98
Range
?Smallest and largest recorded non-missing values.
24 – 578
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.793
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
1.9 / 3.7
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

k_ext_usda_a1065_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

k_ext_usda_a1065_mg_kg distribution02550756 – 23.08: 6423.08 – 40.17: 6640.17 – 57.25: 6257.25 – 74.33: 3474.33 – 91.42: 1491.42 – 108.5: 12108.5 – 125.6: 10125.6 – 142.7: 12142.7 – 159.8: 6159.8 – 176.8: 4176.8 – 193.9: 8193.9 – 211: 0211 – 228.1: 0228.1 – 245.2: 0245.2 – 262.2: 0262.2 – 279.3: 4279.3 – 296.4: 2296.4 – 313.5: 0313.5 – 330.6: 0330.6 – 347.7: 0347.7 – 364.8: 0364.8 – 381.8: 0381.8 – 398.9: 0398.9 – 416: 20200400600
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 2
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
63.73 ± 59.6
Median
?50th percentile: half the non-missing values lie at or below it.
45.5
Range
?Smallest and largest recorded non-missing values.
6 – 416
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.935
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
2.6 / 9.5
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

mg_ext_usda_a1066_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

mg_ext_usda_a1066_mg_kg distribution0501002 – 58.21: 8258.21 – 114.4: 68114.4 – 170.6: 48170.6 – 226.8: 20226.8 – 283: 24283 – 339.2: 18339.2 – 395.5: 10395.5 – 451.7: 6451.7 – 507.9: 4507.9 – 564.1: 4564.1 – 620.3: 0620.3 – 676.5: 6676.5 – 732.7: 4732.7 – 788.9: 0788.9 – 845.1: 0845.1 – 901.3: 2901.3 – 957.5: 0957.5 – 1014: 01014 – 1070: 21070 – 1126: 21126 – 1182: 01182 – 1239: 01239 – 1295: 01295 – 1,351: 205001,0001,500
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
185.2 ± 216
Median
?50th percentile: half the non-missing values lie at or below it.
115
Range
?Smallest and largest recorded non-missing values.
2 – 1,351
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
1.17
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
2.6 / 8.4
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

mn_ext_usda_a1067_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

mn_ext_usda_a1067_mg_kg distribution02550750.7 – 21.92: 6021.92 – 43.13: 6643.13 – 64.35: 3064.35 – 85.57: 1685.57 – 106.8: 26106.8 – 128: 24128 – 149.2: 14149.2 – 170.4: 10170.4 – 191.6: 8191.6 – 212.9: 12212.9 – 234.1: 12234.1 – 255.3: 6255.3 – 276.5: 2276.5 – 297.7: 4297.7 – 318.9: 0318.9 – 340.2: 0340.2 – 361.4: 2361.4 – 382.6: 0382.6 – 403.8: 6403.8 – 425: 0425 – 446.2: 0446.2 – 467.5: 0467.5 – 488.7: 0488.7 – 509.9: 40200400600
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
96.26 ± 98
Median
?50th percentile: half the non-missing values lie at or below it.
59.1
Range
?Smallest and largest recorded non-missing values.
0.7 – 509.9
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
1.02
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
1.8 / 3.7
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

n_tot_usda_a623_w_pct

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

n_tot_usda_a623_w_pct distribution02550750.008 – 0.0255: 540.0255 – 0.043: 740.043 – 0.0605: 520.0605 – 0.078: 280.078 – 0.0955: 300.0955 – 0.113: 260.113 – 0.1305: 60.1305 – 0.148: 80.148 – 0.1655: 60.1655 – 0.183: 00.183 – 0.2005: 40.2005 – 0.218: 20.218 – 0.2355: 40.2355 – 0.253: 00.253 – 0.2705: 20.2705 – 0.288: 00.288 – 0.3055: 00.3055 – 0.323: 20.323 – 0.3405: 00.3405 – 0.358: 00.358 – 0.3755: 00.3755 – 0.393: 00.393 – 0.4105: 20.4105 – 0.428: 20.00.20.40.6
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
0.06857 ± 0.065
Median
?50th percentile: half the non-missing values lie at or below it.
0.051
Range
?Smallest and largest recorded non-missing values.
0.008 – 0.428
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.947
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
2.8 / 10
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

oc_usda_c729_w_pct

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

oc_usda_c729_w_pct distribution02040600.097 – 0.2761: 540.2761 – 0.4552: 600.4552 – 0.6344: 460.6344 – 0.8135: 320.8135 – 0.9926: 200.9926 – 1.172: 241.172 – 1.351: 261.351 – 1.53: 161.53 – 1.709: 41.709 – 1.888: 21.888 – 2.067: 42.067 – 2.246: 22.246 – 2.426: 02.426 – 2.605: 02.605 – 2.784: 22.784 – 2.963: 22.963 – 3.142: 03.142 – 3.321: 23.321 – 3.5: 23.5 – 3.679: 03.679 – 3.859: 03.859 – 4.038: 24.038 – 4.217: 04.217 – 4.396: 20246
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
0.8005 ± 0.709
Median
?50th percentile: half the non-missing values lie at or below it.
0.584
Range
?Smallest and largest recorded non-missing values.
0.097 – 4.396
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.885
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
2.4 / 7.7
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

ph_h2o_usda_a268_index

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

ph_h2o_usda_a268_index distribution020404.1 – 4.276: 44.276 – 4.452: 04.452 – 4.627: 04.627 – 4.803: 144.803 – 4.979: 84.979 – 5.155: 325.155 – 5.331: 285.331 – 5.507: 285.507 – 5.683: 245.683 – 5.858: 125.858 – 6.034: 226.034 – 6.21: 246.21 – 6.386: 306.386 – 6.562: 206.562 – 6.738: 266.738 – 6.913: 86.913 – 7.089: 67.089 – 7.265: 67.265 – 7.441: 27.441 – 7.617: 07.617 – 7.793: 07.793 – 7.968: 07.968 – 8.144: 28.144 – 8.32: 612510
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
5.871 ± 0.773
Median
?50th percentile: half the non-missing values lie at or below it.
5.87
Range
?Smallest and largest recorded non-missing values.
4.1 – 8.32
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.132
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
0.6 / 0.62
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

s_ext_mel3_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

s_ext_mel3_mg_kg distribution01002001 – 4.458: 1844.458 – 7.917: 607.917 – 11.38: 3011.38 – 14.83: 014.83 – 18.29: 818.29 – 21.75: 221.75 – 25.21: 425.21 – 28.67: 228.67 – 32.12: 232.12 – 35.58: 435.58 – 39.04: 239.04 – 42.5: 042.5 – 45.96: 045.96 – 49.42: 049.42 – 52.88: 052.88 – 56.33: 056.33 – 59.79: 259.79 – 63.25: 063.25 – 66.71: 066.71 – 70.17: 070.17 – 73.62: 073.62 – 77.08: 077.08 – 80.54: 080.54 – 84: 20255075100
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
6.344 ± 9.99
Median
?50th percentile: half the non-missing values lie at or below it.
3
Range
?Smallest and largest recorded non-missing values.
1 – 84
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
1.57
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
4.8 / 29
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

zn_ext_usda_a1073_mg_kg

target · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

zn_ext_usda_a1073_mg_kg distribution01002003000.09 – 1.37: 2101.37 – 2.649: 322.649 – 3.929: 123.929 – 5.208: 85.208 – 6.488: 46.488 – 7.768: 07.768 – 9.047: 29.047 – 10.33: 010.33 – 11.61: 011.61 – 12.89: 412.89 – 14.17: 014.17 – 15.45: 215.45 – 16.72: 016.72 – 18: 018 – 19.28: 019.28 – 20.56: 020.56 – 21.84: 021.84 – 23.12: 023.12 – 24.4: 024.4 – 25.68: 025.68 – 26.96: 026.96 – 28.24: 028.24 – 29.52: 029.52 – 30.8: 2010203040
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 26
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
1.526 ± 3.34
Median
?50th percentile: half the non-missing values lie at or below it.
0.5
Range
?Smallest and largest recorded non-missing values.
0.09 – 30.8
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
2.19
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
6 / 45
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

Metadata 7

ID_sample

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

ID_sample classese4003f03c9444aaae892bbd452c50…e4003f03c9444aaae892bbd452c507cd: 22e0e48b077b7409c918eb32cb0f6d1…e0e48b077b7409c918eb32cb0f6d11f7: 22a5545303a66c8698aa1056daa7ca6…a5545303a66c8698aa1056daa7ca6c3c: 2208cdcdfcb4cc359a15a59cbde743e…08cdcdfcb4cc359a15a59cbde743e1c7: 225ff5038ddf8becf15f867dc2dce72…5ff5038ddf8becf15f867dc2dce72bfb: 2207419fd42a72414923d5ed63cc8cd…07419fd42a72414923d5ed63cc8cd6cb: 22e4746156a630b13ddf974d1147e87…e4746156a630b13ddf974d1147e871dd: 22a42fc8c4ee23fcf41de2afb98ad72…a42fc8c4ee23fcf41de2afb98ad72e71: 2230e31e9f37be1670d2b9d83376ffb…30e31e9f37be1670d2b9d83376ffb6be: 226595bec5d54e9c77750a48e740c81…6595bec5d54e9c77750a48e740c81609: 22+10 more+10 more: 2020
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
151
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
e4003f03c9444aaae892bbd452c507cd (2)

collection_name

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

collection_name classesossl_visnirossl_visnir: 151151ossl_mirossl_mir: 151151
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
2
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
ossl_visnir (151)

spectral_kind

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

spectral_kind classesvisnirvisnir: 151151mirmir: 151151
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
2
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
visnir (151)

scan_local_id

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

scan_local_id classesGHBADISTDF3D5DFGGHBADISTDF3D5DFG: 22GHBADISTJT35GNDCGHBADISTJT35GNDC: 22GHBATECS9YMG6O73GHBATECS9YMG6O73: 22WA006775WA006775: 22WA006778WA006778: 22WA006789WA006789: 22WA006802WA006802: 22WA006806WA006806: 22WA006854WA006854: 22WA006904WA006904: 22+10 more+10 more: 2020
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
151
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
GHBADISTDF3D5DFG (2)

feature_count_per_dimension

metadata · numeric

Read the recorded value distribution and summary statistics; no measurement unit is recorded.

feature_count_per_dimension distribution01002001,076 – 1102: 1511102 – 1128: 01128 – 1154: 01154 – 1180: 01180 – 1206: 01206 – 1232: 01232 – 1258: 01258 – 1284: 01284 – 1310: 01310 – 1336: 01336 – 1362: 01362 – 1388: 01388 – 1415: 01415 – 1441: 01441 – 1467: 01467 – 1493: 01493 – 1519: 01519 – 1545: 01545 – 1571: 01571 – 1597: 01597 – 1623: 01623 – 1649: 01649 – 1675: 01675 – 1,701: 1511,0002,0005,00010,000
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Mean ± SD
?Arithmetic mean and standard deviation in the variable's stated unit.
1388 ± 313
Median
?50th percentile: half the non-missing values lie at or below it.
1388
Range
?Smallest and largest recorded non-missing values.
1,076 – 1,701
CV
?Coefficient of variation: standard deviation divided by absolute mean; unstable when the mean is near zero.
0.225
Skew / kurtosis
?Distribution asymmetry and excess kurtosis. A Gaussian distribution has skewness 0 and excess kurtosis 0.
0 / -2
Normal?
?Descriptive normality-test result. A non-normal target is not automatically a defective variable.
no

wavelength_note

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

wavelength_note classesossl_visnir_350_2500_nm_step_2ossl_visnir_350_2500_nm_step_2: 151151ossl_mir_600_4000_cm_minus_1_…ossl_mir_600_4000_cm_minus_1_step_2: 151151
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
2
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
ossl_visnir_350_2500_nm_step_2 (151)

replaces_dataset_id

metadata · categorical

Compare recorded category counts and class balance. Bar lengths show observations in each displayed category.

replaces_dataset_id classesossl_regression_f27317d6ossl_regression_f27317d6: 151151ossl_regression_7b887c0bossl_regression_7b887c0b: 151151
n / missing
?Recorded observation count and number of missing values for this variable.
302 / 0
Classes
?Number of distinct recorded category values.
2
Balance (entropy)
?Normalized Shannon entropy, from 0 for one dominant class to 1 for equal class frequencies.
1
Imbalance ratio
?Largest class count divided by smallest class count; 1 indicates equal counts.
1
Top class
?Most frequent recorded category and its observation count.
ossl_regression_f27317d6 (151)
Constant metadata 10

Fields with one recorded value provide context for every observation; they have no distribution to compare.

  • SpectralRep1
  • datasetOSSL snapshot v1.2
  • dataset_codeAFSIS2.SSL
  • dataset_ownerthe World Agroforestry Centre (ICRAF) and Rothamsted Research (RRES)
  • scan_model_nameBruker Alpha 1 ZnSe
  • scan_model_codeBruker_Alpha_1_ZnSe
  • dimensions1D
  • active_statusactive
  • campaign_date_begin2018-01-01
  • campaign_date_end2018-12-31

4 variable(s) omitted (no recorded values).

Alignment

How observations join to physical sample IDs. Multiple exported rows are not necessarily independent samples or technical repetitions. Reading guide

Alignment levelsample
Sample id availableyes
Samples151
Observations (total)302
Exported observations per unitmin 1 · mean 1 · max 1
Identity statusavailable
Documented technical repetitionsND
Source unitMultiplicity of the declared identifier; protocol not established

Splits

Partitions supplied by the original dataset. They are documented here and are not applied automatically when loading. Reading guide

originalall: 151 documented · not applied

Load this dataset

Use the appropriate route below: an assembled dataset, original raw resources, or the documented access procedure. Reading guide

# Download original raw resources; no Dataverse token required
from nirs4all_datasets import retrieve
result = retrieve("ossl_afsis2_soil_all_y", prepare=False)
print(result["resources"])
# Raw resources may need dataset-specific assembly before analysis

Metadata downloads are available for public datasets only. The dataset bytes are never served here — fetch them from the origin / DOI above.

Read the documentation

Access and rights are separate

Start with provenance and governance at the top of a card. Provenance links to the original resources and publications. Governance records the license, redistribution conditions and content identity.

Open source access means an open original-source route is recorded. Restricted redistribution is the display label for the catalog's private tier: rights are restricted or not cleared for redistribution. It does not mean that the original source is necessarily private.

Raw reproducible identifies a retrieval route. Assembly incomplete means raw resources can be available while dataset-specific conversion is missing. The loading example uses retrieve(id, prepare=False) for these raw downloads; it returns resource metadata rather than X/Y arrays. A complete original-source recipe uses get(id). Two Harvard datasets require explicit guestbook answers.

Token required means no open automatic route is recorded. A Dataverse token works only once the canonical data have a configured hosted DOI. Manual access requires the original provider's documented steps. Availability labels describe the catalog routes, not a fresh live check of every URL.

Why is PCA on a dataset card?

Spectra have many correlated wavelengths. Principal component analysis (PCA) compresses their variation into a few orthogonal directions, called principal components (PCs). It is an exploratory summary of X, calculated without using the target Y.

The score plot places each plotted observation on PC1 and PC2. Nearby dots have similar projections; groups can reveal batches, instruments or natural subpopulations. Axis percentages show explained spectral variance. Large sources are sampled to at most 800 observations. Similar positions in two dimensions do not guarantee similar full spectra; an isolated dot does not prove an acquisition fault.

The scree plot reports spectral variance per component, helping you see whether variation is concentrated in a few dimensions. It is not explained target variance or predictive accuracy. The scree projection and score-plot projection may use different row samples; their percentages need not match.

Q / SPE measures residual variation outside the retained PCA subspace. Hotelling T² measures an observation's scaled distance within that subspace; Mahalanobis H summarizes PCA-space distance. The displayed p95/median ratios compare the distribution's upper tail with its center, without an absolute rejection threshold.

Historical profiles replace missing cells with column medians before PCA, center channels, and use bounded row/component counts. Loadings and distances depend on signal units and preprocessing. Compare compatible sources and inspect the original spectra before acting.

Scores, evidence and missing values

Read measured values first. An alert score maps selected metrics to a heuristic 0–1 scale. The display bands are weak below 0.42, medium from 0.42 to below 0.72, and strong from 0.72. They are not universally validated pass/fail thresholds.

The radar's center is 0 and its outer ring is 1: outward means a stronger anomaly or heterogeneity signal for that axis. Its table gives values rounded to two decimals and explains each axis. Missing axes are left blank; partial profiles have no filled polygon. The mean profile score averages only available axes and is not an overall quality grade.

Diagnostic bars rank possible explanations from heuristic signatures. A score of 0.78 does not mean a 78% causal probability. Evidence coverage is the proportion of required signals available, not confidence in the cause. A dash or “unavailable” means the quantity could not be computed; zero is an actual recorded value.

RMS (root mean square distance) and SAM (Spectral Angle Mapper, in radians) use this dataset's mean spectrum as their reference. They do not compare with an independently acquired white or dark reference. Repeated IDs require a documented protocol before within-ID spread can be called technical repeatability.

X–Y plots show Pearson correlation within this dataset. They identify linear associations; they are not cross-validated predictions and do not establish causal relationships. Targets, metadata, observation counts and physical sample counts have different meanings.

Open the full reading guide