{
  "@context": {
    "@language": "en",
    "@vocab": "https://schema.org/",
    "citeAs": "cr:citeAs",
    "column": "cr:column",
    "conformsTo": "dct:conformsTo",
    "cr": "http://mlcommons.org/croissant/",
    "data": {
      "@id": "cr:data",
      "@type": "@json"
    },
    "dataType": {
      "@id": "cr:dataType",
      "@type": "@vocab"
    },
    "dct": "http://purl.org/dc/terms/",
    "extract": "cr:extract",
    "field": "cr:field",
    "fileObject": "cr:fileObject",
    "fileProperty": "cr:fileProperty",
    "format": "cr:format",
    "includes": "cr:includes",
    "isLiveDataset": "cr:isLiveDataset",
    "key": "cr:key",
    "md5": "cr:md5",
    "parentField": "cr:parentField",
    "path": "cr:path",
    "recordSet": "cr:recordSet",
    "references": "cr:references",
    "repeated": "cr:repeated",
    "sc": "https://schema.org/",
    "sha256": "sha256",
    "source": "cr:source",
    "subField": "cr:subField"
  },
  "@type": "sc:Dataset",
  "citeAs": null,
  "conformsTo": "http://mlcommons.org/croissant/1.0",
  "creator": [],
  "description": "Corn Perten protein NIR. v2.0 standardized NIRS package: 1 spectral source(s), 1 declared target(s). Auto-generated from dataset_card.json (verify before publication).\n\nOpen tier: freely usable and redistributable under the stated license.",
  "distribution": [
    {
      "@id": "file/X.parquet",
      "@type": "cr:FileObject",
      "encodingFormat": "application/vnd.apache.parquet",
      "name": "Corn Perten protein NIR",
      "sha256": "4810af5b9fbaaa93a5f51aae03881090a62cb204b6ce8859b7558051dd6140af"
    },
    {
      "@id": "file/variables.parquet",
      "@type": "cr:FileObject",
      "encodingFormat": "application/vnd.apache.parquet",
      "name": "variables",
      "sha256": "cbdf679c79cbf8f286129fb882f3b6682dbca70f94106f0cbfa58e2958b3bf5d"
    }
  ],
  "keywords": [
    "nir",
    "v2",
    "corn"
  ],
  "license": "CC-BY-4.0",
  "name": "corn_protein_perten",
  "recordSet": [
    {
      "@id": "X",
      "@type": "cr:RecordSet",
      "description": "Spectra from source 'Corn Perten protein NIR' (150 observations x 141 wavelengths).",
      "field": [
        {
          "@id": "X/sample_id",
          "@type": "cr:Field",
          "dataType": "sc:Text",
          "description": "Sample identity (join key across sources and variables).",
          "name": "sample_id",
          "source": {
            "extract": {
              "column": "sample_id"
            },
            "fileObject": {
              "@id": "file/X.parquet"
            }
          }
        },
        {
          "@id": "X/spectrum",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Spectral intensities across the nm axis.",
          "name": "spectrum",
          "repeated": true,
          "source": {
            "fileObject": {
              "@id": "file/X.parquet"
            }
          }
        }
      ],
      "key": {
        "@id": "X/sample_id"
      },
      "name": "X"
    },
    {
      "@id": "variables",
      "@type": "cr:RecordSet",
      "description": "Per-sample target and metadata variables.",
      "field": [
        {
          "@id": "variables/sample_id",
          "@type": "cr:Field",
          "dataType": "sc:Text",
          "description": "Sample identity (join key).",
          "name": "sample_id",
          "source": {
            "extract": {
              "column": "sample_id"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/protein",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'protein' (role target, type numeric).",
          "name": "protein",
          "source": {
            "extract": {
              "column": "protein"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/SpectralRep",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'SpectralRep' (role metadata, type numeric).",
          "name": "SpectralRep",
          "source": {
            "extract": {
              "column": "SpectralRep"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/dataset",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'dataset' (role metadata, type categorical).",
          "name": "dataset",
          "source": {
            "extract": {
              "column": "dataset"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/dataset_title",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'dataset_title' (role metadata, type categorical).",
          "name": "dataset_title",
          "source": {
            "extract": {
              "column": "dataset_title"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/dataset_slug",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'dataset_slug' (role metadata, type categorical).",
          "name": "dataset_slug",
          "source": {
            "extract": {
              "column": "dataset_slug"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/spectroscopy_type",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'spectroscopy_type' (role metadata, type categorical).",
          "name": "spectroscopy_type",
          "source": {
            "extract": {
              "column": "spectroscopy_type"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/cereal_type",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'cereal_type' (role metadata, type categorical).",
          "name": "cereal_type",
          "source": {
            "extract": {
              "column": "cereal_type"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/trait_header",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'trait_header' (role metadata, type categorical).",
          "name": "trait_header",
          "source": {
            "extract": {
              "column": "trait_header"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/trait_header_original",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'trait_header_original' (role metadata, type categorical).",
          "name": "trait_header_original",
          "source": {
            "extract": {
              "column": "trait_header_original"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/task_type",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'task_type' (role metadata, type categorical).",
          "name": "task_type",
          "source": {
            "extract": {
              "column": "task_type"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/raw_label",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'raw_label' (role metadata, type categorical).",
          "name": "raw_label",
          "source": {
            "extract": {
              "column": "raw_label"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/reference_value",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'reference_value' (role metadata, type numeric).",
          "name": "reference_value",
          "source": {
            "extract": {
              "column": "reference_value"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/n_points_original",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'n_points_original' (role metadata, type numeric).",
          "name": "n_points_original",
          "source": {
            "extract": {
              "column": "n_points_original"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/dimensions",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'dimensions' (role metadata, type categorical).",
          "name": "dimensions",
          "source": {
            "extract": {
              "column": "dimensions"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/axis_note",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'axis_note' (role metadata, type categorical).",
          "name": "axis_note",
          "source": {
            "extract": {
              "column": "axis_note"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/ID",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'ID' (role metadata, type categorical).",
          "name": "ID",
          "source": {
            "extract": {
              "column": "ID"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Spectrometer",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'Spectrometer' (role metadata, type categorical).",
          "name": "Spectrometer",
          "source": {
            "extract": {
              "column": "Spectrometer"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Cereal",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'Cereal' (role metadata, type categorical).",
          "name": "Cereal",
          "source": {
            "extract": {
              "column": "Cereal"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Variety",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'Variety' (role metadata, type categorical).",
          "name": "Variety",
          "source": {
            "extract": {
              "column": "Variety"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Country",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'Country' (role metadata, type categorical).",
          "name": "Country",
          "source": {
            "extract": {
              "column": "Country"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Year",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'Year' (role metadata, type numeric).",
          "name": "Year",
          "source": {
            "extract": {
              "column": "Year"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/Moisture",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'Moisture' (role metadata, type numeric).",
          "name": "Moisture",
          "source": {
            "extract": {
              "column": "Moisture"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Spectrometer_Make_Model",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Spectrometer_Make_Model' (role metadata, type categorical).",
          "name": "meta_Spectrometer_Make_Model",
          "source": {
            "extract": {
              "column": "meta_Spectrometer_Make_Model"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Wavelength_Range_nm",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Wavelength_Range_nm' (role metadata, type categorical).",
          "name": "meta_Wavelength_Range_nm",
          "source": {
            "extract": {
              "column": "meta_Wavelength_Range_nm"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Spectral_Resolution_nm",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Spectral_Resolution_nm' (role metadata, type categorical).",
          "name": "meta_Spectral_Resolution_nm",
          "source": {
            "extract": {
              "column": "meta_Spectral_Resolution_nm"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Number_of_Scans_per_Spectrum",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'meta_Number_of_Scans_per_Spectrum' (role metadata, type numeric).",
          "name": "meta_Number_of_Scans_per_Spectrum",
          "source": {
            "extract": {
              "column": "meta_Number_of_Scans_per_Spectrum"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Grain_Type",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Grain_Type' (role metadata, type categorical).",
          "name": "meta_Grain_Type",
          "source": {
            "extract": {
              "column": "meta_Grain_Type"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Variety",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Variety' (role metadata, type categorical).",
          "name": "meta_Variety",
          "source": {
            "extract": {
              "column": "meta_Variety"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Country",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Country' (role metadata, type categorical).",
          "name": "meta_Country",
          "source": {
            "extract": {
              "column": "meta_Country"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Year",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Year' (role metadata, type categorical).",
          "name": "meta_Year",
          "source": {
            "extract": {
              "column": "meta_Year"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Reference_Method_Protein",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Reference_Method_Protein' (role metadata, type categorical).",
          "name": "meta_Reference_Method_Protein",
          "source": {
            "extract": {
              "column": "meta_Reference_Method_Protein"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Reference_Method_Moisture",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Reference_Method_Moisture' (role metadata, type categorical).",
          "name": "meta_Reference_Method_Moisture",
          "source": {
            "extract": {
              "column": "meta_Reference_Method_Moisture"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Units_for_Protein",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Units_for_Protein' (role metadata, type categorical).",
          "name": "meta_Units_for_Protein",
          "source": {
            "extract": {
              "column": "meta_Units_for_Protein"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Units_for_Moisture",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Units_for_Moisture' (role metadata, type categorical).",
          "name": "meta_Units_for_Moisture",
          "source": {
            "extract": {
              "column": "meta_Units_for_Moisture"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Reference_for_moisture",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Reference_for_moisture' (role metadata, type categorical).",
          "name": "meta_Reference_for_moisture",
          "source": {
            "extract": {
              "column": "meta_Reference_for_moisture"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Citation",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Citation' (role metadata, type categorical).",
          "name": "meta_Citation",
          "source": {
            "extract": {
              "column": "meta_Citation"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Autorship",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Autorship' (role metadata, type categorical).",
          "name": "meta_Autorship",
          "source": {
            "extract": {
              "column": "meta_Autorship"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Detector_Type",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Detector_Type' (role metadata, type categorical).",
          "name": "meta_Detector_Type",
          "source": {
            "extract": {
              "column": "meta_Detector_Type"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Geometry",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Geometry' (role metadata, type categorical).",
          "name": "meta_Geometry",
          "source": {
            "extract": {
              "column": "meta_Geometry"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Optical_Reference_Material",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Optical_Reference_Material' (role metadata, type categorical).",
          "name": "meta_Optical_Reference_Material",
          "source": {
            "extract": {
              "column": "meta_Optical_Reference_Material"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Software_Used_for_Acquisition",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Software_Used_for_Acquisition' (role metadata, type categorical).",
          "name": "meta_Software_Used_for_Acquisition",
          "source": {
            "extract": {
              "column": "meta_Software_Used_for_Acquisition"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/meta_Sampling_notes",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'meta_Sampling_notes' (role metadata, type categorical).",
          "name": "meta_Sampling_notes",
          "source": {
            "extract": {
              "column": "meta_Sampling_notes"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/axis_unit",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'axis_unit' (role metadata, type categorical).",
          "name": "axis_unit",
          "source": {
            "extract": {
              "column": "axis_unit"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/axis_min",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'axis_min' (role metadata, type numeric).",
          "name": "axis_min",
          "source": {
            "extract": {
              "column": "axis_min"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/axis_max",
          "@type": "cr:Field",
          "dataType": "sc:Float",
          "description": "Variable 'axis_max' (role metadata, type numeric).",
          "name": "axis_max",
          "source": {
            "extract": {
              "column": "axis_max"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/license",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'license' (role metadata, type categorical).",
          "name": "license",
          "source": {
            "extract": {
              "column": "license"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/license_url",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'license_url' (role metadata, type categorical).",
          "name": "license_url",
          "source": {
            "extract": {
              "column": "license_url"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/rights_status",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'rights_status' (role metadata, type categorical).",
          "name": "rights_status",
          "source": {
            "extract": {
              "column": "rights_status"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/usage_scope",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'usage_scope' (role metadata, type categorical).",
          "name": "usage_scope",
          "source": {
            "extract": {
              "column": "usage_scope"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/public_release_allowed",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'public_release_allowed' (role metadata, type categorical).",
          "name": "public_release_allowed",
          "source": {
            "extract": {
              "column": "public_release_allowed"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/doi_dataset",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'doi_dataset' (role metadata, type categorical).",
          "name": "doi_dataset",
          "source": {
            "extract": {
              "column": "doi_dataset"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        },
        {
          "@id": "variables/notes",
          "@type": "cr:Field",
          "dataType": "sc:Integer",
          "description": "Variable 'notes' (role metadata, type categorical).",
          "name": "notes",
          "source": {
            "extract": {
              "column": "notes"
            },
            "fileObject": {
              "@id": "file/variables.parquet"
            }
          }
        }
      ],
      "key": {
        "@id": "variables/sample_id"
      },
      "name": "variables"
    }
  ],
  "sameAs": [
    "https://zenodo.org/records/15838136",
    "https://zenodo.org/api/records/15838136",
    "10.5281/zenodo.15838136",
    "source_to_standard.py"
  ],
  "url": "https://entrepot.recherche.data.gouv.fr",
  "version": "1.0.0"
}