{
  "acquisition": {
    "acquired_at": "2026-06-27T23:51:23+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "8067fcd7e7a897b6639d9ce28bc09df93b9f8671",
    "source_dataset_id": "fetch_california_housing",
    "source_dataset_name": "California Housing (sklearn)",
    "source_platform": "sklearn",
    "source_rows": 20640,
    "source_target_column_original": "target",
    "source_url": "https://scikit-learn.org/stable/datasets/real_world.html#california-housing-dataset",
    "staged_rows": 1999,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/sklearn-fetch-california-housing-target.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 8,
    "original_columns": 9,
    "original_rows": 1999,
    "rows_after_target_cleanup": 1999,
    "selected_rows": 1999,
    "strata_count": 363,
    "test_rows": 999,
    "train_rows": 1000
  },
  "created_at": "2026-06-27T23:51:23+00:00",
  "dataset_hash": "5f91a00bd3b66742",
  "dedupe_fingerprint": "8067fcd7e7a897b6639d9ce28bc09df93b9f8671",
  "downsampling": {
    "anchors_kept": 18,
    "corr_frob": 0.537513,
    "ks_features_mean": 0.131837,
    "ks_target": 0.015628,
    "method": "hybrid_coreset",
    "minmax_features_preserved": true,
    "minmax_target_preserved": true,
    "n_bins": 50,
    "original_rows": 20640,
    "selected_rows": 1999
  },
  "dropped_feature_columns": [],
  "feature_columns": [
    "MedInc",
    "HouseAge",
    "AveRooms",
    "AveBedrms",
    "Population",
    "AveOccup",
    "Latitude",
    "Longitude"
  ],
  "feature_encodings": {
    "AveBedrms": {
      "impute_value": 1.0782169890664424,
      "kind": "numeric"
    },
    "AveOccup": {
      "impute_value": 2.7481051035876707,
      "kind": "numeric"
    },
    "AveRooms": {
      "impute_value": 5.735359525574499,
      "kind": "numeric"
    },
    "HouseAge": {
      "impute_value": 21.0,
      "kind": "numeric"
    },
    "Latitude": {
      "impute_value": 35.29,
      "kind": "numeric"
    },
    "Longitude": {
      "impute_value": -119.42,
      "kind": "numeric"
    },
    "MedInc": {
      "impute_value": 3.4861,
      "kind": "numeric"
    },
    "Population": {
      "impute_value": 1107.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      15.0001,
      52.0,
      141.9090909090909,
      34.06666666666667,
      35682.0,
      1243.3333333333333,
      41.95,
      -114.31
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      0.4999,
      1.0,
      0.8461538461538461,
      0.3333333333333333,
      3.0,
      0.6923076923076923,
      32.54,
      -124.35
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 8,
  "origin": "real",
  "original_headers": [
    "MedInc",
    "HouseAge",
    "AveRooms",
    "AveBedrms",
    "Population",
    "AveOccup",
    "Latitude",
    "Longitude",
    "target"
  ],
  "output_headers": [
    "MedInc",
    "HouseAge",
    "AveRooms",
    "AveBedrms",
    "Population",
    "AveOccup",
    "Latitude",
    "Longitude",
    "target"
  ],
  "retrieved_at": "2026-06-27T23:51:22+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": 1587840774,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "5f91a00bd3b6674298e103dcebe0f42a057225d3432b5f25deb397720bb17733",
  "source_dataset_id": "fetch_california_housing",
  "source_dataset_name": "California Housing (sklearn)",
  "source_family": "california-housing",
  "source_file": "sklearn-fetch-california-housing-target.csv",
  "source_license": "Public Domain",
  "source_modified_at": "2026-06-27T23:51:23+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/sklearn-fetch-california-housing-target.csv",
  "source_platform": "sklearn",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 20640,
  "source_size_bytes": 181413,
  "source_target_column_original": "target",
  "source_url": "https://scikit-learn.org/stable/datasets/real_world.html#california-housing-dataset",
  "staged_rows": 1999,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/sklearn-fetch-california-housing-target.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 5.00001,
    "min": 0.14999,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 999,
  "train_rows": 1000
}