{
  "acquisition": {
    "acquired_at": "2026-06-27T23:57:35+00:00",
    "dedupe_fingerprint": "3bea935eddd064715ef05d722fe6dc6110018ea6",
    "dropped_feature_columns_for_cap": [],
    "fetched_columns": 9,
    "fetched_rows": 768,
    "openml_data_id": 37,
    "openml_name": "diabetes",
    "row_policy": "kept_all",
    "source_dataset_id": "37",
    "source_dataset_name": "diabetes",
    "source_platform": "openml",
    "source_resource_id": null,
    "source_rows_after_target_cleanup": 768,
    "source_target_column_original": "pres",
    "source_url": "https://www.openml.org/d/37",
    "staged_feature_columns": 8,
    "staged_rows": 768,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/openml-diabetes-37-target-pres.csv",
    "target_profile": {
      "column": "pres",
      "finite_rows": 768,
      "integer_like": true,
      "maximum": 122.0,
      "minimum": 0.0,
      "missing_fraction": 0.0,
      "numeric_valid_fraction": 1.0,
      "unique_values": 47
    },
    "tool": "models.tools.acquire_real_regression_datasets"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 8,
    "original_columns": 9,
    "original_rows": 768,
    "rows_after_target_cleanup": 768,
    "selected_rows": 768,
    "strata_count": 223,
    "test_rows": 128,
    "train_rows": 640
  },
  "created_at": "2026-06-27T23:57:35+00:00",
  "dataset_hash": "f0e7eb3d790df37d",
  "dedupe_fingerprint": "3bea935eddd064715ef05d722fe6dc6110018ea6",
  "dropped_feature_columns": [],
  "feature_columns": [
    "preg",
    "plas",
    "skin",
    "insu",
    "mass",
    "pedi",
    "age",
    "class"
  ],
  "feature_encodings": {
    "age": {
      "impute_value": 29.0,
      "kind": "numeric"
    },
    "class": {
      "impute_value": 0.0,
      "kind": "numeric"
    },
    "insu": {
      "impute_value": 30.5,
      "kind": "numeric"
    },
    "mass": {
      "impute_value": 32.0,
      "kind": "numeric"
    },
    "pedi": {
      "impute_value": 0.3725,
      "kind": "numeric"
    },
    "plas": {
      "impute_value": 117.0,
      "kind": "numeric"
    },
    "preg": {
      "impute_value": 3.0,
      "kind": "numeric"
    },
    "skin": {
      "impute_value": 23.0,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      17.0,
      199.0,
      99.0,
      846.0,
      67.1,
      2.42,
      81.0,
      1.0
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      0.0,
      0.0,
      0.0,
      0.0,
      0.0,
      0.078,
      21.0,
      0.0
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 8,
  "origin": "real",
  "original_headers": [
    "preg",
    "plas",
    "skin",
    "insu",
    "mass",
    "pedi",
    "age",
    "class",
    "target"
  ],
  "output_headers": [
    "preg",
    "plas",
    "skin",
    "insu",
    "mass",
    "pedi",
    "age",
    "class",
    "target"
  ],
  "retrieved_at": "2026-06-27T23:57:35+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": 4057123888,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "f0e7eb3d790df37dfd1e4d52311f37854bdd53b5f1d413bd38d9417e5eed4ee9",
  "source_dataset_id": "37",
  "source_dataset_name": "diabetes",
  "source_family": "diabetes",
  "source_file": "openml-diabetes-37-target-pres.csv",
  "source_license": null,
  "source_modified_at": "2026-06-27T23:57:35+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/openml-diabetes-37-target-pres.csv",
  "source_platform": "openml",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 768,
  "source_size_bytes": 33992,
  "source_target_column_original": "pres",
  "source_url": "https://www.openml.org/d/37",
  "staged_rows": 768,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/openml-diabetes-37-target-pres.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 122.0,
    "min": 0.0,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 128,
  "train_rows": 640
}