{
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 18,
    "original_columns": 19,
    "original_rows": 101,
    "rows_after_target_cleanup": 101,
    "selected_rows": 101,
    "strata_count": 33,
    "test_rows": 17,
    "train_rows": 84
  },
  "created_at": "2026-06-21T07:58:43+00:00",
  "dataset_hash": "03e39a35fac1a204",
  "dropped_feature_columns": [],
  "feature_columns": [
    "pH_target",
    "Clay_target",
    "Altitude",
    "Slope",
    "B01",
    "B02",
    "B03",
    "B04",
    "B05",
    "B06",
    "B07",
    "B08",
    "B8A",
    "B09",
    "B11",
    "B12",
    "NDVI",
    "GNDVI"
  ],
  "feature_encodings": {
    "Altitude": {
      "impute_value": 445.449066162109,
      "kind": "numeric"
    },
    "B01": {
      "impute_value": 1704.0,
      "kind": "numeric"
    },
    "B02": {
      "impute_value": 2034.0,
      "kind": "numeric"
    },
    "B03": {
      "impute_value": 2368.0,
      "kind": "numeric"
    },
    "B04": {
      "impute_value": 2958.0,
      "kind": "numeric"
    },
    "B05": {
      "impute_value": 3257.0,
      "kind": "numeric"
    },
    "B06": {
      "impute_value": 3388.0,
      "kind": "numeric"
    },
    "B07": {
      "impute_value": 3556.0,
      "kind": "numeric"
    },
    "B08": {
      "impute_value": 3630.0,
      "kind": "numeric"
    },
    "B09": {
      "impute_value": 3803.0,
      "kind": "numeric"
    },
    "B11": {
      "impute_value": 5142.0,
      "kind": "numeric"
    },
    "B12": {
      "impute_value": 4200.0,
      "kind": "numeric"
    },
    "B8A": {
      "impute_value": 3783.0,
      "kind": "numeric"
    },
    "Clay_target": {
      "impute_value": 64.5,
      "kind": "numeric"
    },
    "GNDVI": {
      "impute_value": 0.555813953488372,
      "kind": "numeric"
    },
    "NDVI": {
      "impute_value": 0.609903381642512,
      "kind": "numeric"
    },
    "Slope": {
      "impute_value": 0.526809692382812,
      "kind": "numeric"
    },
    "pH_target": {
      "impute_value": 5.67,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      6.41,
      78.7,
      453.252990722656,
      2.42110061645508,
      1750.0,
      2130.0,
      2512.0,
      3204.0,
      3540.0,
      3695.0,
      3922.0,
      3994.0,
      4271.0,
      4181.0,
      5445.0,
      4312.0,
      0.711709545563837,
      0.644486025026313
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      5.07,
      56.0,
      434.226226806641,
      0.121952056884766,
      1556.0,
      1872.0,
      2162.0,
      2730.0,
      3031.0,
      3188.0,
      3351.0,
      3424.0,
      3520.0,
      3586.0,
      4680.0,
      3809.0,
      0.50828729281768,
      0.478172588832487
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 18,
  "original_headers": [
    "pH_target",
    "Clay_target",
    "Altitude",
    "Slope",
    "B01",
    "B02",
    "B03",
    "B04",
    "B05",
    "B06",
    "B07",
    "B08",
    "B8A",
    "B09",
    "B11",
    "B12",
    "NDVI",
    "GNDVI",
    "target"
  ],
  "output_headers": [
    "pH_target",
    "Clay_target",
    "Altitude",
    "Slope",
    "B01",
    "B02",
    "B03",
    "B04",
    "B05",
    "B06",
    "B07",
    "B08",
    "B8A",
    "B09",
    "B11",
    "B12",
    "NDVI",
    "GNDVI",
    "target"
  ],
  "sampling": {
    "method": "quota stratified by target distribution plus high-variance feature quantile bins",
    "minimum_test_to_train_ratio_when_small": 0.2,
    "seed": 47627576,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "03e39a35fac1a204b2ac8899426304af09e780697da2d31da74c9c56bf576722",
  "source_file": "limesoda-mgs-101-dataset.csv",
  "source_modified_at": "2026-06-20T10:07:25+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/limesoda-mgs-101-dataset.csv",
  "source_size_bytes": 17298,
  "target_column": "target",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 2.535,
    "min": 0.791,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 17,
  "train_rows": 84
}