{
  "acquisition": {
    "acquired_at": "2026-06-27T23:53:54+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "6ead0045f8fab76c88fba3b726f698c51780c771",
    "source_dataset_id": "MASS/fgl",
    "source_dataset_name": "Measurements of Forensic Glass Fragments",
    "source_platform": "rdatasets",
    "source_rows": 214,
    "source_target_column_original": "RI",
    "source_url": "https://vincentarelbundock.github.io/Rdatasets/csv/MASS/fgl.csv",
    "staged_rows": 214,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-mass-fgl-ri.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 9,
    "original_columns": 10,
    "original_rows": 214,
    "rows_after_target_cleanup": 214,
    "selected_rows": 214,
    "strata_count": 86,
    "test_rows": 36,
    "train_rows": 178
  },
  "created_at": "2026-06-27T23:53:54+00:00",
  "dataset_hash": "02cc81c6e2492ed0",
  "dedupe_fingerprint": "6ead0045f8fab76c88fba3b726f698c51780c771",
  "dropped_feature_columns": [],
  "feature_columns": [
    "Na",
    "Mg",
    "Al",
    "Si",
    "K",
    "Ca",
    "Ba",
    "Fe",
    "type"
  ],
  "feature_encodings": {
    "Al": {
      "impute_value": 1.36,
      "kind": "numeric"
    },
    "Ba": {
      "impute_value": 0.0,
      "kind": "numeric"
    },
    "Ca": {
      "impute_value": 8.6,
      "kind": "numeric"
    },
    "Fe": {
      "impute_value": 0.0,
      "kind": "numeric"
    },
    "K": {
      "impute_value": 0.555,
      "kind": "numeric"
    },
    "Mg": {
      "impute_value": 3.48,
      "kind": "numeric"
    },
    "Na": {
      "impute_value": 13.3,
      "kind": "numeric"
    },
    "Si": {
      "impute_value": 72.79,
      "kind": "numeric"
    },
    "type": {
      "impute_value": 4.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      17.38,
      4.49,
      3.5,
      75.41,
      6.21,
      16.19,
      3.15,
      0.51,
      5.0
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      10.73,
      0.0,
      0.29,
      69.81,
      0.0,
      5.43,
      0.0,
      0.0,
      0.0
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 9,
  "origin": "real",
  "original_headers": [
    "Na",
    "Mg",
    "Al",
    "Si",
    "K",
    "Ca",
    "Ba",
    "Fe",
    "type",
    "target"
  ],
  "output_headers": [
    "Na",
    "Mg",
    "Al",
    "Si",
    "K",
    "Ca",
    "Ba",
    "Fe",
    "type",
    "target"
  ],
  "retrieved_at": "2026-06-27T23:51:55+00:00",
  "sampling": {
    "final_max_rows": 2000,
    "method": "quota stratified by target distribution plus high-variance feature quantile bins",
    "minimum_final_rows": 101,
    "minimum_test_to_train_ratio_when_small": 0.2,
    "seed": 66692811,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "02cc81c6e2492ed0dd3d364b82e2468fa015b29497816a2d52de4f3f7270c888",
  "source_dataset_id": "MASS/fgl",
  "source_dataset_name": "Measurements of Forensic Glass Fragments",
  "source_family": "measurements-of",
  "source_file": "rdatasets-mass-fgl-ri.csv",
  "source_license": "GPL (R package MASS)",
  "source_modified_at": "2026-06-27T23:53:54+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-mass-fgl-ri.csv",
  "source_platform": "rdatasets",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 214,
  "source_size_bytes": 13360,
  "source_target_column_original": "RI",
  "source_url": "https://vincentarelbundock.github.io/Rdatasets/csv/MASS/fgl.csv",
  "staged_rows": 214,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-mass-fgl-ri.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 15.930000000000064,
    "min": -6.849999999999909,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 36,
  "train_rows": 178
}