{
  "acquisition": {
    "acquired_at": "2026-06-27T22:29:26+00:00",
    "dedupe_fingerprint": "a300ca2eb316a9d58da4fc2f4596868c1b0e4755",
    "fetched_feature_columns": 4,
    "fetched_rows": 559,
    "final_row_cap_policy": "strict_less_than_2000",
    "numeric_target_rows": 559,
    "openml_data_id": 482,
    "openml_name": "arsenic-male-bladder",
    "openml_target_feature": "events",
    "openml_task_id": 4818,
    "openml_task_url": "https://www.openml.org/t/4818",
    "openml_url": "https://www.openml.org/d/482",
    "row_policy": "kept_all",
    "source": "openml",
    "source_dataset_id": "482",
    "source_dataset_name": "arsenic-male-bladder",
    "source_platform": "openml",
    "source_resource_id": "4818",
    "source_rows_after_target_cleanup": 559,
    "source_target_column_original": "events",
    "source_url": "https://www.openml.org/d/482",
    "stage_max_rows": 50000,
    "stage_path": "/home/ubuntu/hyper/data/tabular/realreg-openml-482-events.csv",
    "stage_target_column": "events",
    "staged_columns": 5,
    "staged_feature_columns": 4,
    "staged_rows": 559,
    "staging_path": "/home/ubuntu/hyper/data/tabular/realreg-openml-482-events.csv",
    "target_feature": "events",
    "target_max": 130.0,
    "target_min": 0.0,
    "target_unique_values": 14,
    "tool": "models.tools.acquire_real_regression_datasets"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 4,
    "original_columns": 5,
    "original_rows": 559,
    "rows_after_target_cleanup": 559,
    "selected_rows": 559,
    "strata_count": 27,
    "test_rows": 94,
    "train_rows": 465
  },
  "created_at": "2026-06-27T22:29:26+00:00",
  "dataset_hash": "9b2675742dc80ccd",
  "dedupe_fingerprint": "a300ca2eb316a9d58da4fc2f4596868c1b0e4755",
  "dropped_feature_columns": [],
  "feature_columns": [
    "group",
    "conc",
    "age",
    "at.risk"
  ],
  "feature_encodings": {
    "age": {
      "impute_value": 52.5,
      "kind": "numeric"
    },
    "at.risk": {
      "impute_value": 346.0,
      "kind": "numeric"
    },
    "conc": {
      "impute_value": 259.0,
      "kind": "numeric"
    },
    "group": {
      "impute_value": 22.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      43.0,
      934.0,
      82.5,
      2956638.0
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      1.0,
      0.0,
      22.5,
      10.0
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 4,
  "origin": "real",
  "original_headers": [
    "group",
    "conc",
    "age",
    "at.risk",
    "events"
  ],
  "output_headers": [
    "group",
    "conc",
    "age",
    "at.risk",
    "events"
  ],
  "retrieved_at": "2026-06-27T22:29:26+00:00",
  "sampling": {
    "final_max_rows": 1999,
    "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": 2584957561,
    "source_read_max_rows": null,
    "test_target_rows": 999,
    "train_target_rows": 1000
  },
  "sha256": "9b2675742dc80ccd42db20eaceefc8114bfcbd2555bd01d0c2e8169de0ad2726",
  "source": "openml",
  "source_dataset_id": "482",
  "source_dataset_name": "arsenic-male-bladder",
  "source_file": "realreg-openml-482-events.csv",
  "source_license": null,
  "source_modified_at": "2026-06-27T22:29:26+00:00",
  "source_openml_data_id": 482,
  "source_openml_name": "arsenic-male-bladder",
  "source_openml_task_id": 4818,
  "source_path": "/home/ubuntu/hyper/data/tabular/realreg-openml-482-events.csv",
  "source_platform": "openml",
  "source_real_data": true,
  "source_resource_id": "4818",
  "source_rows": 559,
  "source_size_bytes": 10935,
  "source_target_column_original": "events",
  "source_target_key": "openml:482::events",
  "source_task_url": "https://www.openml.org/t/4818",
  "source_url": "https://www.openml.org/d/482",
  "staged_rows": 559,
  "staging_path": "/home/ubuntu/hyper/data/tabular/realreg-openml-482-events.csv",
  "target_column": "events",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "events"
  ],
  "target_scaling": {
    "max": 130.0,
    "min": 0.0,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 94,
  "train_rows": 465
}