Dataset wisconsin

Feature Scatter

Exoplanet observations derived from the Wisconsin Breast Cancer dataset.

Provenance

{
  "metadata": {
    "id": 1,
    "primary_key": "patient_id",
    "target_variable": "target",
    "original_dataset_type": "medical",
    "obfuscated_dataset_type": "astronomical survey",
    "obfuscated_table_name": "exoplanets",
    "obfuscated_split_table_name": "exoplanet_splits",
    "created_at": "2025-03-07 04:29:39"
  },
  "column_transformations": [
    {
      "original_column": "patient_id",
      "remove": false,
      "obfuscated_column": "exoplanet_id",
      "transformation": null
    },
    {
      "original_column": "mean radius",
      "remove": false,
      "obfuscated_column": "mean_orbital_radius",
      "transformation": "obfuscated['mean_orbital_radius'] = np.log(original['mean radius'] + 1)"
    },
    {
      "original_column": "mean texture",
      "remove": false,
      "obfuscated_column": "mean_surface_roughness",
      "transformation": "obfuscated['mean_surface_roughness'] = original['mean texture'].rank()"
    },
    {
      "original_column": "mean perimeter",
      "remove": false,
      "obfuscated_column": "mean_magnetosphere_extent",
      "transformation": "obfuscated['mean_magnetosphere_extent'] = np.log(original['mean perimeter'] + 1)"
    },
    {
      "original_column": "mean area",
      "remove": false,
      "obfuscated_column": "mean_atmospheric_depth",
      "transformation": "obfuscated['mean_atmospheric_depth'] = np.log(original['mean area'] + 1)"
    },
    {
      "original_column": "mean smoothness",
      "remove": false,
      "obfuscated_column": "mean_tidal_distortion",
      "transformation": "obfuscated['mean_tidal_distortion'] = original['mean smoothness'].rank()"
    },
    {
      "original_column": "mean compactness",
      "remove": false,
      "obfuscated_column": "mean_core_density",
      "transformation": "obfuscated['mean_core_density'] = 1 / (original['mean compactness'] + 1)"
    },
    {
      "original_column": "mean concavity",
      "remove": false,
      "obfuscated_column": "mean_ring_system_complexity",
      "transformation": "obfuscated['mean_ring_system_complexity'] = 1 / (original['mean concavity'] + 1)"
    },
    {
      "original_column": "mean concave points",
      "remove": false,
      "obfuscated_column": "mean_impact_crater_count",
      "transformation": "obfuscated['mean_impact_crater_count'] = 1 / (original['mean concave points'] + 1)"
    },
    {
      "original_column": "mean symmetry",
      "remove": false,
      "obfuscated_column": "mean_axial_symmetry",
      "transformation": "obfuscated['mean_axial_symmetry'] = original['mean symmetry'].rank()"
    },
    {
      "original_column": "mean fractal dimension",
      "remove": false,
      "obfuscated_column": "mean_cloud_turbulence",
      "transformation": "obfuscated['mean_cloud_turbulence'] = original['mean fractal dimension'].rank()"
    },
    {
      "original_column": "radius error",
      "remove": false,
      "obfuscated_column": "orbital_radius_error",
      "transformation": "obfuscated['orbital_radius_error'] = np.sqrt(original['radius error'])"
    },
    {
      "original_column": "texture error",
      "remove": false,
      "obfuscated_column": "surface_roughness_error",
      "transformation": "obfuscated['surface_roughness_error'] = np.sqrt(original['texture error'])"
    },
    {
      "original_column": "perimeter error",
      "remove": false,
      "obfuscated_column": "magnetosphere_extent_error",
      "transformation": "obfuscated['magnetosphere_extent_error'] = np.sqrt(original['perimeter error'])"
    },
    {
      "original_column": "area error",
      "remove": false,
      "obfuscated_column": "atmospheric_depth_error",
      "transformation": "obfuscated['atmospheric_depth_error'] = np.sqrt(original['area error'])"
    },
    {
      "original_column": "smoothness error",
      "remove": false,
      "obfuscated_column": "tidal_distortion_error",
      "transformation": "obfuscated['tidal_distortion_error'] = np.sqrt(original['smoothness error'])"
    },
    {
      "original_column": "compactness error",
      "remove": false,
      "obfuscated_column": "core_density_error",
      "transformation": "obfuscated['core_density_error'] = np.sqrt(original['compactness error'])"
    },
    {
      "original_column": "concavity error",
      "remove": false,
      "obfuscated_column": "ring_system_complexity_error",
      "transformation": "obfuscated['ring_system_complexity_error'] = np.sqrt(original['concavity error'])"
    },
    {
      "original_column": "concave points error",
      "remove": false,
      "obfuscated_column": "impact_crater_count_error",
      "transformation": "obfuscated['impact_crater_count_error'] = np.sqrt(original['concave points error'])"
    },
    {
      "original_column": "symmetry error",
      "remove": false,
      "obfuscated_column": "axial_symmetry_error",
      "transformation": "obfuscated['axial_symmetry_error'] = np.sqrt(original['symmetry error'])"
    },
    {
      "original_column": "fractal dimension error",
      "remove": false,
      "obfuscated_column": "cloud_turbulence_error",
      "transformation": "obfuscated['cloud_turbulence_error'] = np.sqrt(original['fractal dimension error'])"
    },
    {
      "original_column": "worst radius",
      "remove": false,
      "obfuscated_column": "max_orbital_radius",
      "transformation": "obfuscated['max_orbital_radius'] = np.log(original['worst radius'] + 1)"
    },
    {
      "original_column": "worst texture",
      "remove": false,
      "obfuscated_column": "max_surface_roughness",
      "transformation": "obfuscated['max_surface_roughness'] = original['worst texture'].rank()"
    },
    {
      "original_column": "worst perimeter",
      "remove": false,
      "obfuscated_column": "max_magnetosphere_extent",
      "transformation": "obfuscated['max_magnetosphere_extent'] = np.log(original['worst perimeter'] + 1)"
    },
    {
      "original_column": "worst area",
      "remove": false,
      "obfuscated_column": "max_atmospheric_depth",
      "transformation": "obfuscated['max_atmospheric_depth'] = np.log(original['worst area'] + 1)"
    },
    {
      "original_column": "worst smoothness",
      "remove": false,
      "obfuscated_column": "max_tidal_distortion",
      "transformation": "obfuscated['max_tidal_distortion'] = original['worst smoothness'].rank()"
    },
    {
      "original_column": "worst compactness",
      "remove": false,
      "obfuscated_column": "max_core_density",
      "transformation": "obfuscated['max_core_density'] = 1 / (original['worst compactness'] + 1)"
    },
    {
      "original_column": "worst concavity",
      "remove": false,
      "obfuscated_column": "max_ring_system_complexity",
      "transformation": "obfuscated['max_ring_system_complexity'] = 1 / (original['worst concavity'] + 1)"
    },
    {
      "original_column": "worst concave points",
      "remove": false,
      "obfuscated_column": "max_impact_crater_count",
      "transformation": "obfuscated['max_impact_crater_count'] = 1 / (original['worst concave points'] + 1)"
    },
    {
      "original_column": "worst symmetry",
      "remove": false,
      "obfuscated_column": "max_axial_symmetry",
      "transformation": "obfuscated['max_axial_symmetry'] = original['worst symmetry'].rank()"
    },
    {
      "original_column": "worst fractal dimension",
      "remove": false,
      "obfuscated_column": "max_cloud_turbulence",
      "transformation": "obfuscated['max_cloud_turbulence'] = original['worst fractal dimension'].rank()"
    },
    {
      "original_column": "target",
      "remove": false,
      "obfuscated_column": "occupying_species",
      "transformation": "obfuscated['occupying_species'] = original['target'].replace({'malignant': 'Fylaran', 'benign': 'Qtharri'})"
    }
  ]
}

Best Validation Round Rank

Test scores by release date

Regression slope: 0.0007, intercept: -15.0071, p-value: 0.035988

Debug

merge info: 40 rows
filter ollama: 30 rows
drop NA release/accuracy: 30 rows
compute KT and sort
added baselines
create figure
scatter models: [(datetime.date(2024, 1, 25), -0.26330493276166783), (datetime.date(2024, 7, 18), -0.2880874284937826), (datetime.date(2024, 10, 22), -0.07948612109131627), (datetime.date(2024, 10, 22), -0.32813308835098154), (datetime.date(2024, 11, 20), -0.14767309186463937), (datetime.date(2024, 11, 20), -0.120833061346818), (datetime.date(2024, 11, 20), -0.08744657912077478), (datetime.date(2024, 11, 20), -0.06399000689813467), (datetime.date(2024, 12, 5), -0.11224275572500239), (datetime.date(2024, 12, 5), -0.157001482515585), (datetime.date(2024, 12, 5), -0.17628178087575944), (datetime.date(2025, 1, 30), -0.18625270684982762), (datetime.date(2025, 1, 30), -0.157001482515585), (datetime.date(2025, 2, 5), -0.26330493276166783), (datetime.date(2025, 2, 5), -0.22859656282081964), (datetime.date(2025, 2, 19), -0.08744657912077478), (datetime.date(2025, 2, 19), -0.120833061346818), (datetime.date(2025, 2, 27), -0.12959671705266415), (datetime.date(2025, 3, 25), -0.04902781517771927), (datetime.date(2025, 3, 25), -0.04902781517771927), (datetime.date(2025, 4, 14), -0.10381907507268942), (datetime.date(2025, 4, 14), -0.16653464801351997), (datetime.date(2025, 4, 16), -0.1385408635037025), (datetime.date(2025, 4, 16), -0.07948612109131627), (datetime.date(2025, 5, 14), -0.10381907507268942), (datetime.date(2025, 5, 14), -0.07166895281904924), (datetime.date(2025, 5, 14), -0.07166895281904924), (datetime.date(2025, 5, 14), -0.056444480070023374), (datetime.date(2025, 5, 14), -0.04902781517771927), (datetime.date(2025, 5, 14), -0.09555567819126237)]
xlim_min 2024-01-01
scatter ensembles: 6
draw ensemble trend line
xlim_max 2025-05-14
axhline decision trees at -0.070
annotate decision trees at edge 20222.0
mark baseline crossing
savefig website/dataset/wisconsin/release_scores_wisconsin.png

Model Scores

Train vs Validation

Regression slope: 0.9723, intercept: 0.0168, p-value: 2.4894e-13

Validation vs Train

Regression slope: 0.8810, intercept: 0.1012, p-value: 2.4894e-13

Ensemble Max Scores

Baseline

Baseline Models

Logistic Regression

take the atmospheric_depth_error and multiply by -1.829, take the axial_symmetry_error and multiply by 0.024, take the cloud_turbulence_error and multiply by 0.083, take the core_density_error and multiply by 0.309, take the impact_crater_count_error and multiply by -0.003, take the magnetosphere_extent_error and multiply by 0.116, take the max_atmospheric_depth and multiply by -1.690, take the max_axial_symmetry and multiply by -0.007, take the max_cloud_turbulence and multiply by -0.007, take the max_core_density and multiply by 0.048, take the max_impact_crater_count and multiply by 0.290, take the max_magnetosphere_extent and multiply by -0.774, take the max_orbital_radius and multiply by -0.736, take the max_ring_system_complexity and multiply by 0.688, take the max_surface_roughness and multiply by -0.014, take the max_tidal_distortion and multiply by -0.005, take the mean_atmospheric_depth and multiply by -0.989, take the mean_axial_symmetry and multiply by 0.003, take the mean_cloud_turbulence and multiply by 0.009, take the mean_core_density and multiply by -0.019, take the mean_impact_crater_count and multiply by 0.144, take the mean_magnetosphere_extent and multiply by -0.476, take the mean_orbital_radius and multiply by -0.421, take the mean_ring_system_complexity and multiply by 0.295, take the mean_surface_roughness and multiply by -0.000, take the mean_tidal_distortion and multiply by -0.008, take the orbital_radius_error and multiply by 0.019, take the ring_system_complexity_error and multiply by 0.129, take the surface_roughness_error and multiply by 0.931, take the tidal_distortion_error and multiply by 0.063, add those up and if the number is greater than 43.895 then predict Fylaran, otherwise predict Qtharri.

Decision Tree

Dummy

Always predict 1

RuleFit

Bayesian Rule List

CORELS

1. {'col': 'X119', 'index_col': 119, 'cutoff': 0.5, 'val': 0.0, 'flip': True, 'val_right': 0.8034188034188035, 'num_pts': 440, 'num_pts_right': 351}
2. {'val': 0.0, 'num_pts': 89}

EBM