{
  "acquisition": {
    "acquired_at": "2026-06-27T23:58:24+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "71bfa87ec3aabfa406783f0415a152d0f3a5f086",
    "source_dataset_id": "695_chatfield_4",
    "source_dataset_name": "695_chatfield_4",
    "source_platform": "pmlb",
    "source_rows": 235,
    "source_target_column_original": "target",
    "source_url": "https://github.com/EpistasisLab/pmlb/raw/master/datasets/695_chatfield_4/695_chatfield_4.tsv.gz",
    "staged_rows": 235,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-695-chatfield-4-target.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 12,
    "original_columns": 13,
    "original_rows": 235,
    "rows_after_target_cleanup": 235,
    "selected_rows": 235,
    "strata_count": 97,
    "test_rows": 40,
    "train_rows": 195
  },
  "created_at": "2026-06-27T23:58:24+00:00",
  "dataset_hash": "047e62b52a3f4ede",
  "dedupe_fingerprint": "71bfa87ec3aabfa406783f0415a152d0f3a5f086",
  "dropped_feature_columns": [],
  "feature_columns": [
    "col_1",
    "col_2",
    "col_3",
    "col_4",
    "col_5",
    "col_6",
    "col_7",
    "col_8",
    "col_9",
    "col_10",
    "col_11",
    "col_12"
  ],
  "feature_encodings": {
    "col_1": {
      "impute_value": 1866.0,
      "kind": "numeric"
    },
    "col_10": {
      "impute_value": 42.70000076293945,
      "kind": "numeric"
    },
    "col_11": {
      "impute_value": 43.79999923706055,
      "kind": "numeric"
    },
    "col_12": {
      "impute_value": 40.5,
      "kind": "numeric"
    },
    "col_2": {
      "impute_value": 40.59999847412109,
      "kind": "numeric"
    },
    "col_3": {
      "impute_value": 44.0,
      "kind": "numeric"
    },
    "col_4": {
      "impute_value": 45.59999847412109,
      "kind": "numeric"
    },
    "col_5": {
      "impute_value": 41.0,
      "kind": "numeric"
    },
    "col_6": {
      "impute_value": 41.29999923706055,
      "kind": "numeric"
    },
    "col_7": {
      "impute_value": 40.5,
      "kind": "numeric"
    },
    "col_8": {
      "impute_value": 41.900001525878906,
      "kind": "numeric"
    },
    "col_9": {
      "impute_value": 40.70000076293945,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      1983.0,
      217.3999938964844,
      182.3000030517578,
      190.6999969482422,
      196.0,
      238.8999938964844,
      200.6999969482422,
      191.3999938964844,
      200.1999969482422,
      235.8000030517578,
      253.8000030517578,
      210.8999938964844
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      1749.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.0
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 12,
  "origin": "real",
  "original_headers": [
    "col_1",
    "col_2",
    "col_3",
    "col_4",
    "col_5",
    "col_6",
    "col_7",
    "col_8",
    "col_9",
    "col_10",
    "col_11",
    "col_12",
    "target"
  ],
  "output_headers": [
    "col_1",
    "col_2",
    "col_3",
    "col_4",
    "col_5",
    "col_6",
    "col_7",
    "col_8",
    "col_9",
    "col_10",
    "col_11",
    "col_12",
    "target"
  ],
  "retrieved_at": "2026-06-27T23:54:41+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": 88819128,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "047e62b52a3f4ede2f48e684262ce2775bf984d552149be67fd9f99f0067a48b",
  "source_dataset_id": "695_chatfield_4",
  "source_dataset_name": "695_chatfield_4",
  "source_family": "chatfield",
  "source_file": "pmlb-695-chatfield-4-target.csv",
  "source_license": "MIT (PMLB)",
  "source_modified_at": "2026-06-27T23:58:24+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-695-chatfield-4-target.csv",
  "source_platform": "pmlb",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 235,
  "source_size_bytes": 42254,
  "source_target_column_original": "target",
  "source_url": "https://github.com/EpistasisLab/pmlb/raw/master/datasets/695_chatfield_4/695_chatfield_4.tsv.gz",
  "staged_rows": 235,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-695-chatfield-4-target.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 239.3999938964844,
    "min": 0.0,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 40,
  "train_rows": 195
}