{ "analysis_scope": "Held-out comparison of unweighted, non-oversampled logistic regression and random forest, with a dummy-prior baseline, on UCI HTRU2 and UCI Rice", "artifacts": { "fig-held-out-calibration.png": "0fa8d4b3bdbdb590fd3ae3eb03ee6fdb73cfedf244908f6f903593f3b3a6ee20", "fig-held-out-confusion-counts.png": "fcc8c2fba4f1a495d5de0c4d91dd7f058fd98eecb78ae4fddb48d1fde15a5229", "fig-held-out-permutation-importance.png": "6fd583a56de46b5eb3aa2329369c0a14ece9dc6a6478444a3cdd222c270e404b", "fig-held-out-ranking-metrics.png": "a3a3108b2a173162e8471b74b7b9b454c8233816ef6fc6b9de2472a8ab38e9a4", "forest-in-the-sky.ipynb": "dbd80db87a923e10360d9a380b1e066debd8387a4303df2e98f1bf3e91114391", "og-card.png": "1241216e3dcdfad0fd3a2f1da325c24b8f8d0b90a6727d1878958c749ab7d43b", "receipts/calibration-bins.csv": "5f2207281e244b84083b6bb2a19344d90a2c0581453ce56125ace0e5a5cba28c", "receipts/paired-bootstrap.csv": "a228ac24f8a7802831a577fb2320cfb511da937fc0aa2c62f44e21dc6ca0252e", "receipts/permutation-importance.csv": "0db4dc1a13cea576abbd92cdc7c57abafc5722784693c7868301ec02c9704069", "receipts/test-metrics.csv": "acb1a21a822adbc47fb299d8a4a6b04f51700d8b433bf20addf9b87ff1d4c0b8", "receipts/test-predictions.csv": "97203d66565630a1480818fef8f8e6b439b9358cea5fc955a61c06f9f636f0d3", "receipts/training-cv-tuning.csv": "3318a9debe08f495c83a757df5459f9b551e0c75d12a6380db878c8f8e7320ee", "source-manifest.json": "44457cad787a123fbcc8515629d5ac1fc6d9824277355de7e33764121d901a27" }, "datasets": { "htru2": { "calibration": [ { "bin_index": 0, "count": 3580, "dataset": "htru2", "lower_bound_inclusive": 0.0, "mean_predicted_probability": 0.09156306746752339, "model": "dummy_prior", "observed_positive_fraction": 0.09162011173184358, "positive_count": 328, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.1, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.2, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.4, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.5, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.8, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 0, "dataset": "htru2", "lower_bound_inclusive": 0.9, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 1.0, "upper_bound_inclusive": true }, { "bin_index": 0, "count": 3204, "dataset": "htru2", "lower_bound_inclusive": 0.0, "mean_predicted_probability": 0.01232277769947457, "model": "logistic_regression", "observed_positive_fraction": 0.009675405742821473, "positive_count": 31, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 51, "dataset": "htru2", "lower_bound_inclusive": 0.1, "mean_predicted_probability": 0.13629255037494348, "model": "logistic_regression", "observed_positive_fraction": 0.19607843137254902, "positive_count": 10, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 15, "dataset": "htru2", "lower_bound_inclusive": 0.2, "mean_predicted_probability": 0.25018763344294415, "model": "logistic_regression", "observed_positive_fraction": 0.4666666666666667, "positive_count": 7, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 13, "dataset": "htru2", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": 0.3475920944687577, "model": "logistic_regression", "observed_positive_fraction": 0.3076923076923077, "positive_count": 4, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 11, "dataset": "htru2", "lower_bound_inclusive": 0.4, "mean_predicted_probability": 0.44285551780768684, "model": "logistic_regression", "observed_positive_fraction": 0.8181818181818182, "positive_count": 9, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 9, "dataset": "htru2", "lower_bound_inclusive": 0.5, "mean_predicted_probability": 0.5556560288217376, "model": "logistic_regression", "observed_positive_fraction": 0.7777777777777778, "positive_count": 7, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 16, "dataset": "htru2", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": 0.6412765469042143, "model": "logistic_regression", "observed_positive_fraction": 0.75, "positive_count": 12, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 14, "dataset": "htru2", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": 0.7510027649649514, "model": "logistic_regression", "observed_positive_fraction": 0.7857142857142857, "positive_count": 11, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 26, "dataset": "htru2", "lower_bound_inclusive": 0.8, "mean_predicted_probability": 0.8556720930105317, "model": "logistic_regression", "observed_positive_fraction": 0.8076923076923077, "positive_count": 21, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 221, "dataset": "htru2", "lower_bound_inclusive": 0.9, "mean_predicted_probability": 0.9901275582872814, "model": "logistic_regression", "observed_positive_fraction": 0.9773755656108597, "positive_count": 216, "upper_bound": 1.0, "upper_bound_inclusive": true }, { "bin_index": 0, "count": 3189, "dataset": "htru2", "lower_bound_inclusive": 0.0, "mean_predicted_probability": 0.009006756920374406, "model": "random_forest", "observed_positive_fraction": 0.00940733772342427, "positive_count": 30, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 36, "dataset": "htru2", "lower_bound_inclusive": 0.1, "mean_predicted_probability": 0.14578713984530672, "model": "random_forest", "observed_positive_fraction": 0.1388888888888889, "positive_count": 5, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 24, "dataset": "htru2", "lower_bound_inclusive": 0.2, "mean_predicted_probability": 0.2495011576422105, "model": "random_forest", "observed_positive_fraction": 0.125, "positive_count": 3, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 15, "dataset": "htru2", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": 0.35063386522566425, "model": "random_forest", "observed_positive_fraction": 0.5333333333333333, "positive_count": 8, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 12, "dataset": "htru2", "lower_bound_inclusive": 0.4, "mean_predicted_probability": 0.46294546239421536, "model": "random_forest", "observed_positive_fraction": 0.4166666666666667, "positive_count": 5, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 17, "dataset": "htru2", "lower_bound_inclusive": 0.5, "mean_predicted_probability": 0.5556497579211469, "model": "random_forest", "observed_positive_fraction": 0.5294117647058824, "positive_count": 9, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 20, "dataset": "htru2", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": 0.6533116812888291, "model": "random_forest", "observed_positive_fraction": 0.65, "positive_count": 13, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 11, "dataset": "htru2", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": 0.7464046303713053, "model": "random_forest", "observed_positive_fraction": 0.6363636363636364, "positive_count": 7, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 28, "dataset": "htru2", "lower_bound_inclusive": 0.8, "mean_predicted_probability": 0.8623389800361357, "model": "random_forest", "observed_positive_fraction": 0.8571428571428571, "positive_count": 24, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 228, "dataset": "htru2", "lower_bound_inclusive": 0.9, "mean_predicted_probability": 0.9792583224959072, "model": "random_forest", "observed_positive_fraction": 0.9824561403508771, "positive_count": 224, "upper_bound": 1.0, "upper_bound_inclusive": true } ], "dataset": { "audit": { "exact_feature_row_duplicates": 0, "exact_full_row_duplicates": 0, "features": 8, "missing_or_nonfinite_feature_cells": 0, "rows": 17898 }, "class_counts": { "negative": 16259, "positive": 1639 }, "display_name": "HTRU2", "feature_labels": { "dm_snr_excess_kurtosis": "Dispersion-measure signal-to-noise ratio excess kurtosis", "dm_snr_mean": "Dispersion-measure signal-to-noise ratio mean", "dm_snr_skewness": "Dispersion-measure signal-to-noise ratio skewness", "dm_snr_standard_deviation": "Dispersion-measure signal-to-noise ratio standard deviation", "profile_excess_kurtosis": "Profile excess kurtosis", "profile_mean": "Profile mean", "profile_skewness": "Profile skewness", "profile_standard_deviation": "Profile standard deviation" }, "features": [ "profile_mean", "profile_standard_deviation", "profile_excess_kurtosis", "profile_skewness", "dm_snr_mean", "dm_snr_standard_deviation", "dm_snr_excess_kurtosis", "dm_snr_skewness" ], "key": "htru2", "negative_label": "non-pulsar candidate", "positive_label": "pulsar candidate", "positive_prevalence": 0.091574, "row_definition": "one human-annotated pulsar candidate feature row", "source_path": "data/HTRU_2.csv", "source_sha256": "b13b4d8929e96ecd196e464c1c8a454c3ac2ffa631015f6388957531a9923f59" }, "paired_bootstrap": { "confidence_level_label": "95% bootstrap stability interval", "interval_endpoints": { "lower_percentile": 2.5, "quantile_method": "numpy.quantile(method='linear')", "upper_percentile": 97.5 }, "limitations": [ "conditions on the single fixed test split and its observed class counts", "does not include training-split, tuning, refitting, or forest-seed variation", "assumes held-out rows are exchangeable within class", "does not repair unrecorded grouping or dependence in either public dataset", "does not establish performance in a new survey, field, instrument, or population" ], "method": "paired stratified nonparametric bootstrap of the fixed held-out rows", "models": { "dummy_prior": { "average_precision": { "lower": 0.09162, "median": 0.09162, "upper": 0.09162 }, "balanced_accuracy": { "lower": 0.5, "median": 0.5, "upper": 0.5 }, "brier_score": { "lower": 0.083226, "median": 0.083226, "upper": 0.083226 }, "f1_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "log_loss": { "lower": 0.30627, "median": 0.30627, "upper": 0.30627 }, "precision_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "recall_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "roc_auc": { "lower": 0.5, "median": 0.5, "upper": 0.5 } }, "logistic_regression": { "average_precision": { "lower": 0.885479, "median": 0.915733, "upper": 0.939644 }, "balanced_accuracy": { "lower": 0.883049, "median": 0.904847, "upper": 0.924817 }, "brier_score": { "lower": 0.015059, "median": 0.018336, "upper": 0.021835 }, "f1_at_0_5": { "lower": 0.841762, "median": 0.870279, "upper": 0.896774 }, "log_loss": { "lower": 0.064444, "median": 0.077703, "upper": 0.093162 }, "precision_at_0_5": { "lower": 0.905259, "median": 0.934256, "upper": 0.96058 }, "recall_at_0_5": { "lower": 0.771341, "median": 0.814024, "upper": 0.856707 }, "roc_auc": { "lower": 0.964824, "median": 0.976012, "upper": 0.985212 } }, "random_forest": { "average_precision": { "lower": 0.903718, "median": 0.927143, "upper": 0.948268 }, "balanced_accuracy": { "lower": 0.898749, "median": 0.918707, "upper": 0.93717 }, "brier_score": { "lower": 0.014243, "median": 0.017391, "upper": 0.020804 }, "f1_at_0_5": { "lower": 0.850153, "median": 0.877358, "upper": 0.903021 }, "log_loss": { "lower": 0.059326, "median": 0.071933, "upper": 0.086106 }, "precision_at_0_5": { "lower": 0.880114, "median": 0.911765, "upper": 0.941383 }, "recall_at_0_5": { "lower": 0.804878, "median": 0.844512, "upper": 0.881098 }, "roc_auc": { "lower": 0.966554, "median": 0.976945, "upper": 0.986439 } } }, "not_a_population_confidence_interval": true, "random_forest_minus_logistic_regression": { "delta_definition": "random forest metric minus logistic regression metric", "interpretation": "positive favors random forest for higher-is-better metrics; negative favors random forest for lower-is-better log loss and Brier score", "metrics": { "average_precision": { "lower": 0.001192, "median": 0.011187, "upper": 0.025677 }, "balanced_accuracy": { "lower": 0.003651, "median": 0.01386, "upper": 0.025453 }, "brier_score": { "lower": -0.0022, "median": -0.00095, "upper": 0.000286 }, "f1_at_0_5": { "lower": -0.009359, "median": 0.007095, "upper": 0.023848 }, "log_loss": { "lower": -0.011639, "median": -0.005723, "upper": -0.000456 }, "precision_at_0_5": { "lower": -0.044632, "median": -0.022234, "upper": -0.00213 }, "recall_at_0_5": { "lower": 0.009146, "median": 0.030488, "upper": 0.051905 }, "roc_auc": { "lower": -0.00309, "median": 0.001034, "upper": 0.005399 } } }, "random_state": 20301093, "replicates": 2000, "resampling": "positive and negative held-out row indices are sampled with replacement within class; the same sampled indices are used for every model" }, "permutation_importance": { "computed_on": "fixed held-out test rows after model selection", "limitations": [ "repeat spread is not a confidence interval", "correlated features can share or mask importance", "permutation can create feature combinations not common in the source table" ], "models": { "logistic_regression": [ { "feature": "profile_excess_kurtosis", "feature_label": "Profile excess kurtosis", "mean_average_precision_decrease": 0.856496, "standard_deviation_across_permutations": 0.002016 }, { "feature": "dm_snr_standard_deviation", "feature_label": "Dispersion-measure signal-to-noise ratio standard deviation", "mean_average_precision_decrease": 0.08032, "standard_deviation_across_permutations": 0.005833 }, { "feature": "profile_skewness", "feature_label": "Profile skewness", "mean_average_precision_decrease": 0.018759, "standard_deviation_across_permutations": 0.006379 }, { "feature": "dm_snr_mean", "feature_label": "Dispersion-measure signal-to-noise ratio mean", "mean_average_precision_decrease": 0.017582, "standard_deviation_across_permutations": 0.002827 }, { "feature": "dm_snr_skewness", "feature_label": "Dispersion-measure signal-to-noise ratio skewness", "mean_average_precision_decrease": 0.008555, "standard_deviation_across_permutations": 0.002194 }, { "feature": "profile_mean", "feature_label": "Profile mean", "mean_average_precision_decrease": 0.003887, "standard_deviation_across_permutations": 0.002777 }, { "feature": "dm_snr_excess_kurtosis", "feature_label": "Dispersion-measure signal-to-noise ratio excess kurtosis", "mean_average_precision_decrease": 0.003165, "standard_deviation_across_permutations": 0.002135 }, { "feature": "profile_standard_deviation", "feature_label": "Profile standard deviation", "mean_average_precision_decrease": 0.000258, "standard_deviation_across_permutations": 0.001195 } ], "random_forest": [ { "feature": "profile_excess_kurtosis", "feature_label": "Profile excess kurtosis", "mean_average_precision_decrease": 0.659815, "standard_deviation_across_permutations": 0.014306 }, { "feature": "dm_snr_standard_deviation", "feature_label": "Dispersion-measure signal-to-noise ratio standard deviation", "mean_average_precision_decrease": 0.018536, "standard_deviation_across_permutations": 0.00371 }, { "feature": "profile_skewness", "feature_label": "Profile skewness", "mean_average_precision_decrease": 0.015324, "standard_deviation_across_permutations": 0.005422 }, { "feature": "profile_mean", "feature_label": "Profile mean", "mean_average_precision_decrease": 0.007959, "standard_deviation_across_permutations": 0.001703 }, { "feature": "profile_standard_deviation", "feature_label": "Profile standard deviation", "mean_average_precision_decrease": 0.006573, "standard_deviation_across_permutations": 0.002531 }, { "feature": "dm_snr_excess_kurtosis", "feature_label": "Dispersion-measure signal-to-noise ratio excess kurtosis", "mean_average_precision_decrease": 0.004943, "standard_deviation_across_permutations": 0.001471 }, { "feature": "dm_snr_mean", "feature_label": "Dispersion-measure signal-to-noise ratio mean", "mean_average_precision_decrease": 0.004602, "standard_deviation_across_permutations": 0.002031 }, { "feature": "dm_snr_skewness", "feature_label": "Dispersion-measure signal-to-noise ratio skewness", "mean_average_precision_decrease": 0.004335, "standard_deviation_across_permutations": 0.002515 } ] }, "not_used_for_tuning": true, "random_states": { "logistic_regression": 20311093, "random_forest": 20311094 }, "repeats": 30, "scoring": "average_precision" }, "scope_limits": [ "the public table provides no grouping identifiers for dependence-aware splitting", "a fixed row-level holdout does not establish performance in a new acquisition context", "test-set interpretation does not turn association into causation", "model comparison is conditional on the declared grids, split, threshold, and metrics" ], "split": { "method": "fixed stratified random holdout", "random_state": 20261093, "test_fraction": 0.2, "test_original_index_sha256": "25257ea11745e4c8ae1cafa492cb2142f2d79b1296f6d0f122137fc987279c3b", "test_positives": 328, "test_rows": 3580, "training_original_index_sha256": "a98590af7a3c53ffdb0b484a00025e1e3ae52dcc08ec662565f6c135dd87c319", "training_positives": 1311, "training_rows": 14318 }, "test_metrics": { "dummy_prior": { "average_precision": 0.09162011173184358, "balanced_accuracy": 0.5, "brier_score": 0.08322587011213615, "confusion": { "fn": 328, "fp": 0, "tn": 3252, "tp": 0 }, "f1_at_0_5": 0.0, "log_loss": 0.306270251420438, "precision_at_0_5": 0.0, "recall_at_0_5": 0.0, "roc_auc": 0.5 }, "logistic_regression": { "average_precision": 0.9141203134286622, "balanced_accuracy": 0.904090915909159, "brier_score": 0.01842103175623111, "confusion": { "fn": 61, "fp": 19, "tn": 3233, "tp": 267 }, "f1_at_0_5": 0.8697068403908794, "log_loss": 0.07838934833153892, "precision_at_0_5": 0.9335664335664335, "recall_at_0_5": 0.8140243902439024, "roc_auc": 0.9756238187381874 }, "random_forest": { "average_precision": 0.926209011132564, "balanced_accuracy": 0.9181048060480604, "brier_score": 0.01748606782101274, "confusion": { "fn": 51, "fp": 27, "tn": 3225, "tp": 277 }, "f1_at_0_5": 0.8765822784810128, "log_loss": 0.07250472858171678, "precision_at_0_5": 0.9111842105263158, "recall_at_0_5": 0.8445121951219512, "roc_auc": 0.9766836730867308 } }, "training_only_tuning": { "folds": 5, "logistic_regression": { "best_mean_validation_average_precision": 0.928258, "best_params": { "classifier__C": 1.0 }, "class_weight": null, "converged": true, "fixed_parameters": { "max_iter": 5000, "penalty": "l2", "random_state": 20281093, "solver": "lbfgs" }, "grid": { "C": [ 0.01, 0.1, 1.0, 10.0, 100.0 ] }, "oversampling": null, "pipeline": [ "StandardScaler", "LogisticRegression" ], "selected_fit_iterations": [ 20 ] }, "objective": "average_precision", "random_forest": { "best_mean_validation_average_precision": 0.930227, "best_params": { "max_depth": 6, "max_features": 0.75, "min_samples_leaf": 5 }, "class_weight": null, "fixed_parameters": { "bootstrap": true, "criterion": "gini", "n_jobs": 1, "random_state": 20291093 }, "grid": { "max_depth": [ 6, 12, null ], "max_features": [ "sqrt", 0.75 ], "min_samples_leaf": [ 1, 5 ] }, "n_estimators": 250, "oversampling": null }, "random_state": 20271093, "selection_tie_behavior": "GridSearchCV selects the first candidate in the declared ParameterGrid order among exactly tied mean validation scores", "shared_fold_records": [ { "fold": 0, "training_original_index_sha256": "dfddd22802dd00568a284759a0c21f200f622639bb0e9b15b150053b45c99667", "training_rows": 11454, "validation_original_index_sha256": "eca05d6505bfe9a1a9153e5b60ffeb6d785bcdc7499262cde616a9523f2a64c3", "validation_positives": 262, "validation_rows": 2864 }, { "fold": 1, "training_original_index_sha256": "45569109dad1e46c6a6cfd607d4c0200371db770390dbc45f7a8133ddda96a11", "training_rows": 11454, "validation_original_index_sha256": "ea088db4bef9c18cd9b2073e5c5e88a0a69e6cd771b20c5385a78bc2ef1ee864", "validation_positives": 262, "validation_rows": 2864 }, { "fold": 2, "training_original_index_sha256": "01ed6e32566a7480eb71f206a38685a5670328604ba67f20925361dd04a18665", "training_rows": 11454, "validation_original_index_sha256": "833cf16ecb33c009034ed7e40ce99cc898f2f7b44504d071507ed35779c0e5b6", "validation_positives": 263, "validation_rows": 2864 }, { "fold": 3, "training_original_index_sha256": "827f089a1a53b4d83e4633974898009fd48943b2459ec557c213571e020d3398", "training_rows": 11455, "validation_original_index_sha256": "8f36e684f070aacd69c33cfa8a22a803a5e37ac5e2d725f0380566f8eeea1e63", "validation_positives": 262, "validation_rows": 2863 }, { "fold": 4, "training_original_index_sha256": "21a7d58bc1f4e0b76bedeed467e187e3669aa1a106e6cc8927d5d21ddda55faf", "training_rows": 11455, "validation_original_index_sha256": "fe51299d2b439502eacf9e842a087ab588a1b2019d9a54a44c7f8e9c6b8d3dea", "validation_positives": 262, "validation_rows": 2863 } ], "test_set_role": "kept outside fitting and hyperparameter tuning within the published workflow; used for descriptive evaluation and post-evaluation diagnostics" } }, "rice": { "calibration": [ { "bin_index": 0, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.0, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.1, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.2, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 762, "dataset": "rice", "lower_bound_inclusive": 0.4, "mean_predicted_probability": 0.42782152230971127, "model": "dummy_prior", "observed_positive_fraction": 0.42782152230971127, "positive_count": 326, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.5, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.8, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 0, "dataset": "rice", "lower_bound_inclusive": 0.9, "mean_predicted_probability": null, "model": "dummy_prior", "observed_positive_fraction": null, "positive_count": 0, "upper_bound": 1.0, "upper_bound_inclusive": true }, { "bin_index": 0, "count": 370, "dataset": "rice", "lower_bound_inclusive": 0.0, "mean_predicted_probability": 0.015946994709722606, "model": "logistic_regression", "observed_positive_fraction": 0.04054054054054054, "positive_count": 15, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 28, "dataset": "rice", "lower_bound_inclusive": 0.1, "mean_predicted_probability": 0.14966149288431432, "model": "logistic_regression", "observed_positive_fraction": 0.14285714285714285, "positive_count": 4, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 19, "dataset": "rice", "lower_bound_inclusive": 0.2, "mean_predicted_probability": 0.2516632767912769, "model": "logistic_regression", "observed_positive_fraction": 0.3157894736842105, "positive_count": 6, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 14, "dataset": "rice", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": 0.35903989370385564, "model": "logistic_regression", "observed_positive_fraction": 0.35714285714285715, "positive_count": 5, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 16, "dataset": "rice", "lower_bound_inclusive": 0.4, "mean_predicted_probability": 0.44506823165717246, "model": "logistic_regression", "observed_positive_fraction": 0.5625, "positive_count": 9, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 15, "dataset": "rice", "lower_bound_inclusive": 0.5, "mean_predicted_probability": 0.5348772322042196, "model": "logistic_regression", "observed_positive_fraction": 0.4666666666666667, "positive_count": 7, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 14, "dataset": "rice", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": 0.6427163502109202, "model": "logistic_regression", "observed_positive_fraction": 0.7142857142857143, "positive_count": 10, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 22, "dataset": "rice", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": 0.7513054846588166, "model": "logistic_regression", "observed_positive_fraction": 0.8181818181818182, "positive_count": 18, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 31, "dataset": "rice", "lower_bound_inclusive": 0.8, "mean_predicted_probability": 0.8534312351004558, "model": "logistic_regression", "observed_positive_fraction": 0.8387096774193549, "positive_count": 26, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 233, "dataset": "rice", "lower_bound_inclusive": 0.9, "mean_predicted_probability": 0.9808135302629741, "model": "logistic_regression", "observed_positive_fraction": 0.9699570815450643, "positive_count": 226, "upper_bound": 1.0, "upper_bound_inclusive": true }, { "bin_index": 0, "count": 347, "dataset": "rice", "lower_bound_inclusive": 0.0, "mean_predicted_probability": 0.012937287197469633, "model": "random_forest", "observed_positive_fraction": 0.03170028818443804, "positive_count": 11, "upper_bound": 0.1, "upper_bound_inclusive": false }, { "bin_index": 1, "count": 49, "dataset": "rice", "lower_bound_inclusive": 0.1, "mean_predicted_probability": 0.1324746291062678, "model": "random_forest", "observed_positive_fraction": 0.20408163265306123, "positive_count": 10, "upper_bound": 0.2, "upper_bound_inclusive": false }, { "bin_index": 2, "count": 17, "dataset": "rice", "lower_bound_inclusive": 0.2, "mean_predicted_probability": 0.24357033713689963, "model": "random_forest", "observed_positive_fraction": 0.29411764705882354, "positive_count": 5, "upper_bound": 0.30000000000000004, "upper_bound_inclusive": false }, { "bin_index": 3, "count": 18, "dataset": "rice", "lower_bound_inclusive": 0.30000000000000004, "mean_predicted_probability": 0.34880905069444634, "model": "random_forest", "observed_positive_fraction": 0.2222222222222222, "positive_count": 4, "upper_bound": 0.4, "upper_bound_inclusive": false }, { "bin_index": 4, "count": 19, "dataset": "rice", "lower_bound_inclusive": 0.4, "mean_predicted_probability": 0.44916902894608873, "model": "random_forest", "observed_positive_fraction": 0.5789473684210527, "positive_count": 11, "upper_bound": 0.5, "upper_bound_inclusive": false }, { "bin_index": 5, "count": 10, "dataset": "rice", "lower_bound_inclusive": 0.5, "mean_predicted_probability": 0.5529158860742089, "model": "random_forest", "observed_positive_fraction": 0.8, "positive_count": 8, "upper_bound": 0.6000000000000001, "upper_bound_inclusive": false }, { "bin_index": 6, "count": 14, "dataset": "rice", "lower_bound_inclusive": 0.6000000000000001, "mean_predicted_probability": 0.6550448952748208, "model": "random_forest", "observed_positive_fraction": 0.7142857142857143, "positive_count": 10, "upper_bound": 0.7000000000000001, "upper_bound_inclusive": false }, { "bin_index": 7, "count": 32, "dataset": "rice", "lower_bound_inclusive": 0.7000000000000001, "mean_predicted_probability": 0.7597709181295642, "model": "random_forest", "observed_positive_fraction": 0.65625, "positive_count": 21, "upper_bound": 0.8, "upper_bound_inclusive": false }, { "bin_index": 8, "count": 24, "dataset": "rice", "lower_bound_inclusive": 0.8, "mean_predicted_probability": 0.8601326951494492, "model": "random_forest", "observed_positive_fraction": 0.8333333333333334, "positive_count": 20, "upper_bound": 0.9, "upper_bound_inclusive": false }, { "bin_index": 9, "count": 232, "dataset": "rice", "lower_bound_inclusive": 0.9, "mean_predicted_probability": 0.9782588919039424, "model": "random_forest", "observed_positive_fraction": 0.9741379310344828, "positive_count": 226, "upper_bound": 1.0, "upper_bound_inclusive": true } ], "dataset": { "audit": { "exact_feature_row_duplicates": 0, "exact_full_row_duplicates": 0, "features": 7, "missing_or_nonfinite_feature_cells": 0, "rows": 3810 }, "class_counts": { "negative": 2180, "positive": 1630 }, "display_name": "Rice", "feature_labels": { "area": "Area", "convex_area": "Convex area", "eccentricity": "Eccentricity", "extent": "Extent", "major_axis_length": "Major axis length", "minor_axis_length": "Minor axis length", "perimeter": "Perimeter" }, "features": [ "area", "perimeter", "major_axis_length", "minor_axis_length", "eccentricity", "convex_area", "extent" ], "key": "rice", "negative_label": "Osmancik", "positive_label": "Cammeo", "positive_prevalence": 0.427822, "row_definition": "one segmented rice-grain image feature row", "source_path": "data/Rice_Cammeo_Osmancik.arff", "source_sha256": "1af97883100c89de2ea2972f7a28d428f4f1c14711a61defc0b0569e9eb65665" }, "paired_bootstrap": { "confidence_level_label": "95% bootstrap stability interval", "interval_endpoints": { "lower_percentile": 2.5, "quantile_method": "numpy.quantile(method='linear')", "upper_percentile": 97.5 }, "limitations": [ "conditions on the single fixed test split and its observed class counts", "does not include training-split, tuning, refitting, or forest-seed variation", "assumes held-out rows are exchangeable within class", "does not repair unrecorded grouping or dependence in either public dataset", "does not establish performance in a new survey, field, instrument, or population" ], "method": "paired stratified nonparametric bootstrap of the fixed held-out rows", "models": { "dummy_prior": { "average_precision": { "lower": 0.427822, "median": 0.427822, "upper": 0.427822 }, "balanced_accuracy": { "lower": 0.5, "median": 0.5, "upper": 0.5 }, "brier_score": { "lower": 0.24479, "median": 0.24479, "upper": 0.24479 }, "f1_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "log_loss": { "lower": 0.682691, "median": 0.682691, "upper": 0.682691 }, "precision_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "recall_at_0_5": { "lower": 0.0, "median": 0.0, "upper": 0.0 }, "roc_auc": { "lower": 0.5, "median": 0.5, "upper": 0.5 } }, "logistic_regression": { "average_precision": { "lower": 0.951994, "median": 0.964661, "upper": 0.97549 }, "balanced_accuracy": { "lower": 0.887002, "median": 0.908088, "upper": 0.928772 }, "brier_score": { "lower": 0.052317, "median": 0.065163, "upper": 0.079294 }, "f1_at_0_5": { "lower": 0.870813, "median": 0.895801, "upper": 0.919088 }, "log_loss": { "lower": 0.181939, "median": 0.22569, "upper": 0.275062 }, "precision_at_0_5": { "lower": 0.881615, "median": 0.91195, "upper": 0.940204 }, "recall_at_0_5": { "lower": 0.843558, "median": 0.880368, "upper": 0.91411 }, "roc_auc": { "lower": 0.95701, "median": 0.969075, "upper": 0.979218 } }, "random_forest": { "average_precision": { "lower": 0.947273, "median": 0.961087, "upper": 0.972723 }, "balanced_accuracy": { "lower": 0.885483, "median": 0.906913, "upper": 0.927211 }, "brier_score": { "lower": 0.054828, "median": 0.068091, "upper": 0.082256 }, "f1_at_0_5": { "lower": 0.869427, "median": 0.894231, "upper": 0.917574 }, "log_loss": { "lower": 0.193138, "median": 0.241228, "upper": 0.294893 }, "precision_at_0_5": { "lower": 0.884239, "median": 0.914474, "upper": 0.942182 }, "recall_at_0_5": { "lower": 0.840491, "median": 0.874233, "upper": 0.911043 }, "roc_auc": { "lower": 0.95156, "median": 0.965, "upper": 0.97666 } } }, "not_a_population_confidence_interval": true, "random_forest_minus_logistic_regression": { "delta_definition": "random forest metric minus logistic regression metric", "interpretation": "positive favors random forest for higher-is-better metrics; negative favors random forest for lower-is-better log loss and Brier score", "metrics": { "average_precision": { "lower": -0.007324, "median": -0.003554, "upper": 0.000217 }, "balanced_accuracy": { "lower": -0.011869, "median": -0.001534, "upper": 0.008429 }, "brier_score": { "lower": -0.000151, "median": 0.002873, "upper": 0.005927 }, "f1_at_0_5": { "lower": -0.013527, "median": -0.001714, "upper": 0.0101 }, "log_loss": { "lower": 0.003355, "median": 0.015493, "upper": 0.027459 }, "precision_at_0_5": { "lower": -0.011906, "median": 0.002553, "upper": 0.018371 }, "recall_at_0_5": { "lower": -0.021472, "median": -0.006135, "upper": 0.009202 }, "roc_auc": { "lower": -0.006965, "median": -0.004037, "upper": -0.001147 } } }, "random_state": 20301266, "replicates": 2000, "resampling": "positive and negative held-out row indices are sampled with replacement within class; the same sampled indices are used for every model" }, "permutation_importance": { "computed_on": "fixed held-out test rows after model selection", "limitations": [ "repeat spread is not a confidence interval", "correlated features can share or mask importance", "permutation can create feature combinations not common in the source table" ], "models": { "logistic_regression": [ { "feature": "convex_area", "feature_label": "Convex area", "mean_average_precision_decrease": 0.650849, "standard_deviation_across_permutations": 0.00458 }, { "feature": "major_axis_length", "feature_label": "Major axis length", "mean_average_precision_decrease": 0.251018, "standard_deviation_across_permutations": 0.014859 }, { "feature": "area", "feature_label": "Area", "mean_average_precision_decrease": 0.102356, "standard_deviation_across_permutations": 0.008537 }, { "feature": "perimeter", "feature_label": "Perimeter", "mean_average_precision_decrease": 0.032271, "standard_deviation_across_permutations": 0.004638 }, { "feature": "minor_axis_length", "feature_label": "Minor axis length", "mean_average_precision_decrease": 0.022697, "standard_deviation_across_permutations": 0.003648 }, { "feature": "eccentricity", "feature_label": "Eccentricity", "mean_average_precision_decrease": 0.010375, "standard_deviation_across_permutations": 0.002867 }, { "feature": "extent", "feature_label": "Extent", "mean_average_precision_decrease": 0.000112, "standard_deviation_across_permutations": 7.7e-05 } ], "random_forest": [ { "feature": "major_axis_length", "feature_label": "Major axis length", "mean_average_precision_decrease": 0.073807, "standard_deviation_across_permutations": 0.010075 }, { "feature": "perimeter", "feature_label": "Perimeter", "mean_average_precision_decrease": 0.039134, "standard_deviation_across_permutations": 0.007802 }, { "feature": "convex_area", "feature_label": "Convex area", "mean_average_precision_decrease": 0.017065, "standard_deviation_across_permutations": 0.004499 }, { "feature": "eccentricity", "feature_label": "Eccentricity", "mean_average_precision_decrease": 0.009946, "standard_deviation_across_permutations": 0.00286 }, { "feature": "area", "feature_label": "Area", "mean_average_precision_decrease": 0.008566, "standard_deviation_across_permutations": 0.003814 }, { "feature": "minor_axis_length", "feature_label": "Minor axis length", "mean_average_precision_decrease": 0.000959, "standard_deviation_across_permutations": 0.00074 }, { "feature": "extent", "feature_label": "Extent", "mean_average_precision_decrease": 0.000131, "standard_deviation_across_permutations": 0.000525 } ] }, "not_used_for_tuning": true, "random_states": { "logistic_regression": 20311266, "random_forest": 20311267 }, "repeats": 30, "scoring": "average_precision" }, "scope_limits": [ "the public table provides no grouping identifiers for dependence-aware splitting", "a fixed row-level holdout does not establish performance in a new acquisition context", "test-set interpretation does not turn association into causation", "model comparison is conditional on the declared grids, split, threshold, and metrics" ], "split": { "method": "fixed stratified random holdout", "random_state": 20261266, "test_fraction": 0.2, "test_original_index_sha256": "757bba843e140089f520a51abe705db10fe9d3b39ddb9d2ddec20d61cf7b815b", "test_positives": 326, "test_rows": 762, "training_original_index_sha256": "894d4b13deff0f77c169869567f5283535e89e0968495e091dcd860ef500898e", "training_positives": 1304, "training_rows": 3048 }, "test_metrics": { "dummy_prior": { "average_precision": 0.42782152230971127, "balanced_accuracy": 0.5, "brier_score": 0.2447902673583125, "confusion": { "fn": 326, "fp": 0, "tn": 436, "tp": 0 }, "f1_at_0_5": 0.0, "log_loss": 0.6826912217946177, "precision_at_0_5": 0.0, "recall_at_0_5": 0.0, "roc_auc": 0.5 }, "logistic_regression": { "average_precision": 0.9642983404255487, "balanced_accuracy": 0.9080739573366354, "brier_score": 0.06551801677560717, "confusion": { "fn": 39, "fp": 28, "tn": 408, "tp": 287 }, "f1_at_0_5": 0.8954758190327614, "log_loss": 0.22644837545370158, "precision_at_0_5": 0.9111111111111111, "recall_at_0_5": 0.8803680981595092, "roc_auc": 0.9688326673045533 }, "random_forest": { "average_precision": 0.9607270950794979, "balanced_accuracy": 0.9061532616648843, "brier_score": 0.06844252554461035, "confusion": { "fn": 41, "fp": 27, "tn": 409, "tp": 285 }, "f1_at_0_5": 0.8934169278996864, "log_loss": 0.2419898205497901, "precision_at_0_5": 0.9134615384615384, "recall_at_0_5": 0.8742331288343558, "roc_auc": 0.964790763775539 } }, "training_only_tuning": { "folds": 5, "logistic_regression": { "best_mean_validation_average_precision": 0.977383, "best_params": { "classifier__C": 100.0 }, "class_weight": null, "converged": true, "fixed_parameters": { "max_iter": 5000, "penalty": "l2", "random_state": 20281266, "solver": "lbfgs" }, "grid": { "C": [ 0.01, 0.1, 1.0, 10.0, 100.0 ] }, "oversampling": null, "pipeline": [ "StandardScaler", "LogisticRegression" ], "selected_fit_iterations": [ 31 ] }, "objective": "average_precision", "random_forest": { "best_mean_validation_average_precision": 0.97529, "best_params": { "max_depth": 6, "max_features": "sqrt", "min_samples_leaf": 5 }, "class_weight": null, "fixed_parameters": { "bootstrap": true, "criterion": "gini", "n_jobs": 1, "random_state": 20291266 }, "grid": { "max_depth": [ 6, 12, null ], "max_features": [ "sqrt", 0.75 ], "min_samples_leaf": [ 1, 5 ] }, "n_estimators": 250, "oversampling": null }, "random_state": 20271266, "selection_tie_behavior": "GridSearchCV selects the first candidate in the declared ParameterGrid order among exactly tied mean validation scores", "shared_fold_records": [ { "fold": 0, "training_original_index_sha256": "fa890207af32e7672d9d7fe6c5be8155abcceccd09e0f3796128cf9a1b66f291", "training_rows": 2438, "validation_original_index_sha256": "d8290b63e1784e8f10f57eb680e8833e0d3e1307d1bcd040d53e433cfc5f0d4f", "validation_positives": 261, "validation_rows": 610 }, { "fold": 1, "training_original_index_sha256": "8a0d9d89b11344fa406f0d801118f28a9bd9f908c8a487ff276dcae11821b268", "training_rows": 2438, "validation_original_index_sha256": "355a07873584f8d1ba74f691e180233a2ae421e9b99eae76172fd273a2cdb84a", "validation_positives": 261, "validation_rows": 610 }, { "fold": 2, "training_original_index_sha256": "4f3a884d960bb1a4eae4a4381196a192e03a0e4ef7218d0819cf50c2b8ac7235", "training_rows": 2438, "validation_original_index_sha256": "2ef371eeb4f08e93d97f64f63b010496aa6a5e6de970c1803e7ac4a14921c936", "validation_positives": 261, "validation_rows": 610 }, { "fold": 3, "training_original_index_sha256": "aa2910f6b6f63a0ed00a127ed41b5aa6f000b0e728cdb652aac88cf46429c311", "training_rows": 2439, "validation_original_index_sha256": "38a0abcccd0655c128ed8520703027237a2dec4c5a53e6d0f1fc5c20884c20c9", "validation_positives": 261, "validation_rows": 609 }, { "fold": 4, "training_original_index_sha256": "222e0a81f3f680abf92dbd1e2585e53b4de00e32e880086a42bffb17079e63e7", "training_rows": 2439, "validation_original_index_sha256": "111fd87cb66b855a44406c9894983336fee78e2c4b5b6b12b3e07b537cb800ac", "validation_positives": 260, "validation_rows": 609 } ], "test_set_role": "kept outside fitting and hyperparameter tuning within the published workflow; used for descriptive evaluation and post-evaluation diagnostics" } } }, "protocol": { "class_weighting": null, "model_comparison": "same source rows, split, training folds, and test metrics within each dataset", "oversampling": null, "primary_metric": "average_precision", "selection_history": "dataset, model, and narrative choices followed exploratory screening of these public datasets; held-out estimates are descriptive, not a prospective confirmation", "split": "fixed stratified 80/20 holdout created before learned preprocessing", "test_rows_outside_fitting_and_tuning_in_published_workflow": true, "threshold_metrics": "fixed probability threshold 0.5", "tuning": "five shared stratified training-only folds per dataset" }, "provenance": { "attribution": "ATTRIBUTION.txt", "attribution_sha256": "09160838d8093cbaf05d47b99a08c45065fdd830a362c27a1a99a305951d5d21", "generator": "reproduce.py", "generator_sha256": "a2ad91d1e3f7cf2a15da4fc10246de57b7adda70c1ab2f229f2e51553a75e510", "lockfile": "reproduce.py.lock", "lockfile_sha256": "004d05f217f53bb88abdddb954262afa464daa0366fddb3d35fc11b65e82cfb9", "reference_environment": { "architecture": "arm64", "figure_font": "DejaVuSans.ttf", "figure_font_sha256": "3fdf69cabf06049ea70a00b5919340e2ce1e6d02b0cc3c4b44fb6801bd1e0d22", "freetype_version": "2.6.1", "matplotlib_backend": "Agg", "matplotlib_version": "3.10.3", "numpy_version": "2.3.2", "operating_system": "Darwin", "os_release": "25.5.0", "python_implementation": "CPython", "python_version": "3.12.12", "scikit_learn_version": "1.7.1" }, "source_hashes": { "data/Citation_Request.txt": "7184131b2ca0a456f619ede4cec7cf84607f8d9b32e6f5aab5be6e59f52dfc48", "data/HTRU_2.csv": "b13b4d8929e96ecd196e464c1c8a454c3ac2ffa631015f6388957531a9923f59", "data/Readme.txt": "691efe1b5b910401959a9b4f74ed0959dcd205d69c73cf524646e6f63a3eb86b", "data/Rice_Cammeo_Osmancik.arff": "1af97883100c89de2ea2972f7a28d428f4f1c14711a61defc0b0569e9eb65665" } }, "runtime_note": "elapsed_seconds fields in training diagnostics are informational and are not publication artifacts because wall time is not byte-reproducible", "schema_version": 1, "study_id": "uci-htru2-rice-logistic-random-forest-v1", "verification_scope": { "claim": "byte identity in the recorded reference environment only", "cross_platform_note": "operating systems, numerical libraries, FreeType builds, and font files can change output bytes; canonical release artifacts should be generated and verified in the same pinned Linux environment", "environment_recorded": true } }