{
  "acquisition": {
    "acquired_at": "2026-06-28T00:13:57+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "52f3f4ed49dc1ff2a75348aa79213cdf1f0aefa1",
    "source_dataset_id": "Ecdat/CPSch3",
    "source_dataset_name": "Earnings from the Current Population Survey",
    "source_platform": "rdatasets",
    "source_rows": 11130,
    "source_target_column_original": "ahe",
    "source_url": "https://vincentarelbundock.github.io/Rdatasets/csv/Ecdat/CPSch3.csv",
    "staged_rows": 1999,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-ecdat-cpsch3-ahe.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 2,
    "original_columns": 3,
    "original_rows": 1999,
    "rows_after_target_cleanup": 1999,
    "selected_rows": 1999,
    "strata_count": 24,
    "test_rows": 999,
    "train_rows": 1000
  },
  "created_at": "2026-06-28T00:13:57+00:00",
  "dataset_hash": "1434fbab3d10ab05",
  "dedupe_fingerprint": "52f3f4ed49dc1ff2a75348aa79213cdf1f0aefa1",
  "downsampling": {
    "anchors_kept": 5,
    "corr_frob": 0.71823,
    "ks_features_mean": 0.469556,
    "ks_target": 0.467776,
    "method": "hybrid_coreset",
    "minmax_features_preserved": true,
    "minmax_target_preserved": true,
    "n_bins": 50,
    "original_rows": 11130,
    "selected_rows": 1999
  },
  "dropped_feature_columns": [],
  "feature_columns": [
    "year",
    "sex"
  ],
  "feature_encodings": {
    "sex": {
      "impute_value": 1.0,
      "kind": "numeric"
    },
    "year": {
      "impute_value": 1992.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      1998.0,
      1.0
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      1992.0,
      0.0
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 2,
  "origin": "real",
  "original_headers": [
    "year",
    "sex",
    "target"
  ],
  "output_headers": [
    "year",
    "sex",
    "target"
  ],
  "retrieved_at": "2026-06-28T00:12:15+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": 352443558,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "1434fbab3d10ab0506c31424a7b6e6dca06863dbb68b277ffb1ed8014614a625",
  "source_dataset_id": "Ecdat/CPSch3",
  "source_dataset_name": "Earnings from the Current Population Survey",
  "source_family": "earnings-from",
  "source_file": "rdatasets-ecdat-cpsch3-ahe.csv",
  "source_license": "GPL (R package Ecdat)",
  "source_modified_at": "2026-06-28T00:13:57+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-ecdat-cpsch3-ahe.csv",
  "source_platform": "rdatasets",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 11130,
  "source_size_bytes": 55856,
  "source_target_column_original": "ahe",
  "source_url": "https://vincentarelbundock.github.io/Rdatasets/csv/Ecdat/CPSch3.csv",
  "staged_rows": 1999,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/rdatasets-ecdat-cpsch3-ahe.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 52.4433746337891,
    "min": 2.13648986816406,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 999,
  "train_rows": 1000
}