{
  "acquisition": {
    "acquired_at": "2026-06-27T23:58:05+00:00",
    "agent_id": "Claude-extra",
    "dedupe_fingerprint": "4ae9be14477743f88c1df962f89de0a82d2d06c1",
    "source_dataset_id": "583_fri_c1_1000_50",
    "source_dataset_name": "583_fri_c1_1000_50",
    "source_platform": "pmlb",
    "source_rows": 1000,
    "source_target_column_original": "target",
    "source_url": "https://github.com/EpistasisLab/pmlb/raw/master/datasets/583_fri_c1_1000_50/583_fri_c1_1000_50.tsv.gz",
    "staged_rows": 1000,
    "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-583-fri-c1-1000-50-target.csv",
    "tool": "models.tools.acquire_regression_extra_sources"
  },
  "counts": {
    "dropped_feature_count": 0,
    "n_features": 50,
    "original_columns": 51,
    "original_rows": 1000,
    "rows_after_target_cleanup": 1000,
    "selected_rows": 1000,
    "strata_count": 159,
    "test_rows": 167,
    "train_rows": 833
  },
  "created_at": "2026-06-27T23:58:05+00:00",
  "dataset_hash": "b78d323566930bd7",
  "dedupe_fingerprint": "4ae9be14477743f88c1df962f89de0a82d2d06c1",
  "dropped_feature_columns": [],
  "feature_columns": [
    "oz1",
    "oz2",
    "oz3",
    "oz4",
    "oz5",
    "oz6",
    "oz7",
    "oz8",
    "oz9",
    "oz10",
    "oz11",
    "oz12",
    "oz13",
    "oz14",
    "oz15",
    "oz16",
    "oz17",
    "oz18",
    "oz19",
    "oz20",
    "oz21",
    "oz22",
    "oz23",
    "oz24",
    "oz25",
    "oz26",
    "oz27",
    "oz28",
    "oz29",
    "oz30",
    "oz31",
    "oz32",
    "oz33",
    "oz34",
    "oz35",
    "oz36",
    "oz37",
    "oz38",
    "oz39",
    "oz40",
    "oz41",
    "oz42",
    "oz43",
    "oz44",
    "oz45",
    "oz46",
    "oz47",
    "oz48",
    "oz49",
    "oz50"
  ],
  "feature_encodings": {
    "oz1": {
      "impute_value": -0.028391172178089598,
      "kind": "numeric"
    },
    "oz10": {
      "impute_value": 0.0175499124452471,
      "kind": "numeric"
    },
    "oz11": {
      "impute_value": 0.0042441970435902,
      "kind": "numeric"
    },
    "oz12": {
      "impute_value": -0.035514076240360695,
      "kind": "numeric"
    },
    "oz13": {
      "impute_value": -0.0018246195395477,
      "kind": "numeric"
    },
    "oz14": {
      "impute_value": 0.008788362611085149,
      "kind": "numeric"
    },
    "oz15": {
      "impute_value": 0.0017228511860594002,
      "kind": "numeric"
    },
    "oz16": {
      "impute_value": 0.0239720540121197,
      "kind": "numeric"
    },
    "oz17": {
      "impute_value": 0.039397930726408945,
      "kind": "numeric"
    },
    "oz18": {
      "impute_value": 0.013845055364072299,
      "kind": "numeric"
    },
    "oz19": {
      "impute_value": 0.0072238696739076996,
      "kind": "numeric"
    },
    "oz2": {
      "impute_value": -0.04831439256668085,
      "kind": "numeric"
    },
    "oz20": {
      "impute_value": -0.0009263103129342499,
      "kind": "numeric"
    },
    "oz21": {
      "impute_value": 0.0519417896866798,
      "kind": "numeric"
    },
    "oz22": {
      "impute_value": -0.0127467075362801,
      "kind": "numeric"
    },
    "oz23": {
      "impute_value": 0.0079283600207418,
      "kind": "numeric"
    },
    "oz24": {
      "impute_value": 0.0229091020300984,
      "kind": "numeric"
    },
    "oz25": {
      "impute_value": -0.02028003241866825,
      "kind": "numeric"
    },
    "oz26": {
      "impute_value": -0.01842891145497555,
      "kind": "numeric"
    },
    "oz27": {
      "impute_value": -0.022099956870079,
      "kind": "numeric"
    },
    "oz28": {
      "impute_value": -0.0785675793886184,
      "kind": "numeric"
    },
    "oz29": {
      "impute_value": 0.02301626745611425,
      "kind": "numeric"
    },
    "oz3": {
      "impute_value": -0.0294413855299353,
      "kind": "numeric"
    },
    "oz30": {
      "impute_value": 0.00038551428588105,
      "kind": "numeric"
    },
    "oz31": {
      "impute_value": 0.01495096739381545,
      "kind": "numeric"
    },
    "oz32": {
      "impute_value": -0.014567393809556949,
      "kind": "numeric"
    },
    "oz33": {
      "impute_value": 0.0426364336162805,
      "kind": "numeric"
    },
    "oz34": {
      "impute_value": 0.0142292981036007,
      "kind": "numeric"
    },
    "oz35": {
      "impute_value": -0.0720685645937919,
      "kind": "numeric"
    },
    "oz36": {
      "impute_value": 0.019932166673243,
      "kind": "numeric"
    },
    "oz37": {
      "impute_value": 0.034957587718963595,
      "kind": "numeric"
    },
    "oz38": {
      "impute_value": 0.0182423116639256,
      "kind": "numeric"
    },
    "oz39": {
      "impute_value": -0.036402428522706004,
      "kind": "numeric"
    },
    "oz4": {
      "impute_value": 0.0364765170961618,
      "kind": "numeric"
    },
    "oz40": {
      "impute_value": -0.03524104785174125,
      "kind": "numeric"
    },
    "oz41": {
      "impute_value": 0.012127044610679101,
      "kind": "numeric"
    },
    "oz42": {
      "impute_value": 0.047214690595865194,
      "kind": "numeric"
    },
    "oz43": {
      "impute_value": 0.0108126793056726,
      "kind": "numeric"
    },
    "oz44": {
      "impute_value": -0.00339421103126365,
      "kind": "numeric"
    },
    "oz45": {
      "impute_value": -0.0035072108148597,
      "kind": "numeric"
    },
    "oz46": {
      "impute_value": 0.012015559244900852,
      "kind": "numeric"
    },
    "oz47": {
      "impute_value": -0.0212800092995166,
      "kind": "numeric"
    },
    "oz48": {
      "impute_value": -0.0133750620298087,
      "kind": "numeric"
    },
    "oz49": {
      "impute_value": 0.0032492654863744497,
      "kind": "numeric"
    },
    "oz5": {
      "impute_value": 0.04488222673535345,
      "kind": "numeric"
    },
    "oz50": {
      "impute_value": 0.036353401839733096,
      "kind": "numeric"
    },
    "oz6": {
      "impute_value": 0.035961342975497204,
      "kind": "numeric"
    },
    "oz7": {
      "impute_value": -0.0146351968869566,
      "kind": "numeric"
    },
    "oz8": {
      "impute_value": -0.00030362850520759994,
      "kind": "numeric"
    },
    "oz9": {
      "impute_value": 0.03267696313560005,
      "kind": "numeric"
    }
  },
  "feature_scaling": {
    "constant_columns": [],
    "max": [
      2.4428369998931885,
      1.774325609207153,
      1.720121145248413,
      1.676460027694702,
      1.6973495483398438,
      1.6809659004211426,
      1.8165186643600464,
      1.8042895793914795,
      1.7092219591140747,
      1.6841826438903809,
      1.8583042621612549,
      1.7442104816436768,
      1.70269513130188,
      1.690056562423706,
      1.6909106969833374,
      1.7097282409667969,
      1.647365689277649,
      1.735440731048584,
      1.661786675453186,
      1.690913200378418,
      1.700456619262695,
      1.6962268352508545,
      1.740193247795105,
      1.7549967765808103,
      1.733178734779358,
      1.794360637664795,
      1.781273365020752,
      1.7734777927398682,
      1.666049242019653,
      1.7206857204437256,
      1.7519575357437134,
      1.749252200126648,
      1.703311800956726,
      1.7162538766860962,
      1.784982681274414,
      1.725631833076477,
      1.7270681858062744,
      1.6882801055908203,
      1.7135082483291626,
      1.6654798984527588,
      1.7331626415252686,
      1.77069091796875,
      1.794636845588684,
      1.754507541656494,
      1.691866159439087,
      1.7215652465820312,
      1.7420203685760498,
      1.7303074598312378,
      1.7135326862335205,
      1.678494215011597
    ],
    "method": "minmax over cleaned source rows before sampling",
    "min": [
      -2.5003278255462646,
      -1.6981569528579712,
      -1.714128851890564,
      -1.7757353782653809,
      -1.7111868858337402,
      -1.7899080514907837,
      -1.7809288501739502,
      -1.6839812994003296,
      -1.758044958114624,
      -1.7407981157302856,
      -1.7423181533813477,
      -1.802329540252685,
      -1.7392345666885376,
      -1.7593835592269895,
      -1.7228225469589231,
      -1.7892301082611084,
      -1.7703607082366943,
      -1.771018147468567,
      -1.7014849185943604,
      -1.6979413032531738,
      -1.7798938751220703,
      -1.7169853448867798,
      -1.7517404556274414,
      -1.6909830570220947,
      -1.726301193237305,
      -1.720003962516785,
      -1.7261512279510498,
      -1.6806166172027588,
      -1.7063556909561155,
      -1.6910537481307983,
      -1.725653886795044,
      -1.7222338914871216,
      -1.733340859413147,
      -1.744644045829773,
      -1.621485471725464,
      -1.6779508590698242,
      -1.810908555984497,
      -1.7726200819015503,
      -1.641684651374817,
      -1.7538896799087524,
      -1.7284095287322998,
      -1.78120756149292,
      -1.7312462329864502,
      -1.7274701595306396,
      -1.7183233499526978,
      -1.8342920541763303,
      -1.7030518054962158,
      -1.7706283330917358,
      -1.70911967754364,
      -1.7213212251663208
    ],
    "range": [
      0.0,
      1.0
    ]
  },
  "hash_algorithm": "sha256:first16",
  "headers_removed_from_csv": true,
  "n_features": 50,
  "origin": "real",
  "original_headers": [
    "oz1",
    "oz2",
    "oz3",
    "oz4",
    "oz5",
    "oz6",
    "oz7",
    "oz8",
    "oz9",
    "oz10",
    "oz11",
    "oz12",
    "oz13",
    "oz14",
    "oz15",
    "oz16",
    "oz17",
    "oz18",
    "oz19",
    "oz20",
    "oz21",
    "oz22",
    "oz23",
    "oz24",
    "oz25",
    "oz26",
    "oz27",
    "oz28",
    "oz29",
    "oz30",
    "oz31",
    "oz32",
    "oz33",
    "oz34",
    "oz35",
    "oz36",
    "oz37",
    "oz38",
    "oz39",
    "oz40",
    "oz41",
    "oz42",
    "oz43",
    "oz44",
    "oz45",
    "oz46",
    "oz47",
    "oz48",
    "oz49",
    "oz50",
    "target"
  ],
  "output_headers": [
    "oz1",
    "oz2",
    "oz3",
    "oz4",
    "oz5",
    "oz6",
    "oz7",
    "oz8",
    "oz9",
    "oz10",
    "oz11",
    "oz12",
    "oz13",
    "oz14",
    "oz15",
    "oz16",
    "oz17",
    "oz18",
    "oz19",
    "oz20",
    "oz21",
    "oz22",
    "oz23",
    "oz24",
    "oz25",
    "oz26",
    "oz27",
    "oz28",
    "oz29",
    "oz30",
    "oz31",
    "oz32",
    "oz33",
    "oz34",
    "oz35",
    "oz36",
    "oz37",
    "oz38",
    "oz39",
    "oz40",
    "oz41",
    "oz42",
    "oz43",
    "oz44",
    "oz45",
    "oz46",
    "oz47",
    "oz48",
    "oz49",
    "oz50",
    "target"
  ],
  "retrieved_at": "2026-06-27T23:54:41+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": 3065517368,
    "source_read_max_rows": null,
    "test_target_rows": 1000,
    "train_target_rows": 1000
  },
  "sha256": "b78d323566930bd7c6842bd688964491e586aa611af8ec96158c9290ff46c784",
  "source_dataset_id": "583_fri_c1_1000_50",
  "source_dataset_name": "583_fri_c1_1000_50",
  "source_family": "fri-c1",
  "source_file": "pmlb-583-fri-c1-1000-50-target.csv",
  "source_license": "MIT (PMLB)",
  "source_modified_at": "2026-06-27T23:58:05+00:00",
  "source_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-583-fri-c1-1000-50-target.csv",
  "source_platform": "pmlb",
  "source_real_data": true,
  "source_resource_id": null,
  "source_rows": 1000,
  "source_size_bytes": 984050,
  "source_target_column_original": "target",
  "source_url": "https://github.com/EpistasisLab/pmlb/raw/master/datasets/583_fri_c1_1000_50/583_fri_c1_1000_50.tsv.gz",
  "staged_rows": 1000,
  "staging_path": "/home/ubuntu/hyper/data/tabular/real_regression_sources/pmlb-583-fri-c1-1000-50-target.csv",
  "target_column": "target",
  "target_generation": "original_numeric_column",
  "target_is_final_column": true,
  "target_names": [
    "target"
  ],
  "target_scaling": {
    "max": 2.216630697250366,
    "min": -2.642117500305176,
    "mode": "regression",
    "range": [
      0.0,
      1.0
    ]
  },
  "target_source_encoding": {
    "kind": "numeric"
  },
  "task_type": "regression",
  "test_rows": 167,
  "train_rows": 833
}