{
  "acquisition": {
    "acquired_at": "2026-06-27T23:51:24+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "93b0ecbeb3a2ef4359fbea34b1a3d45ab4ca6e71",
    "source_dataset_id": "healthexp",
    "source_dataset_name": "Health expenditure by country",
    "source_platform": "seaborn",
    "source_rows": 274,
    "source_target_column_original": "Spending_USD",
    "source_url": "https://github.com/mwaskom/seaborn-data/blob/master/healthexp.csv",
    "staged_rows": 274,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/seaborn-healthexp-spending-usd.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 3,
    "original_columns": 4,
    "original_rows": 274,
    "rows_after_target_cleanup": 274,
    "selected_rows": 274,
    "strata_count": 67,
    "test_rows": 46,
    "train_rows": 228
  },
  "created_at": "2026-06-27T23:51:24+00:00",
  "dataset_hash": "3f781a51af45387e",
  "dedupe_fingerprint": "93b0ecbeb3a2ef4359fbea34b1a3d45ab4ca6e71",
  "dropped_feature_columns": [],
  "feature_columns": [
    "Year",
    "Country",
    "Life_Expectancy"
  ],
  "feature_encodings": {
    "Country": {
      "impute_value": 3.0,
      "kind": "numeric"
    },
    "Life_Expectancy": {
      "impute_value": 78.1,
      "kind": "numeric"
    },
    "Year": {
      "impute_value": 1998.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      2020.0,
      5.0,
      84.7
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      1970.0,
      0.0,
      70.6
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 3,
  "origin": "real",
  "original_headers": [
    "Year",
    "Country",
    "Life_Expectancy",
    "target"
  ],
  "output_headers": [
    "Year",
    "Country",
    "Life_Expectancy",
    "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": 1045249372,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "3f781a51af45387ee3598e45797720e9e780a3ef16186dd50efa08b23083925d",
  "source_dataset_id": "healthexp",
  "source_dataset_name": "Health expenditure by country",
  "source_family": "health-expenditure",
  "source_file": "seaborn-healthexp-spending-usd.csv",
  "source_license": "CC0 (seaborn-data)",
  "source_modified_at": "2026-06-27T23:51:24+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/seaborn-healthexp-spending-usd.csv",
  "source_platform": "seaborn",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 274,
  "source_size_bytes": 6795,
  "source_target_column_original": "Spending_USD",
  "source_url": "https://github.com/mwaskom/seaborn-data/blob/master/healthexp.csv",
  "staged_rows": 274,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/seaborn-healthexp-spending-usd.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 11859.179,
    "min": 123.993,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 46,
  "train_rows": 228
}