.. DO NOT EDIT.
.. THIS FILE WAS AUTOMATICALLY GENERATED BY SPHINX-GALLERY.
.. TO MAKE CHANGES, EDIT THE SOURCE PYTHON FILE:
.. "auto_examples/snsx/plot_decileplot_script.py"
.. LINE NUMBERS ARE GIVEN BELOW.

.. only:: html

    .. note::
        :class: sphx-glr-download-link-note

        :ref:`Go to the end <sphx_glr_download_auto_examples_snsx_plot_decileplot_script.py>`
        to download the full example code. or to run this example in your browser via JupyterLite or Binder

.. rst-class:: sphx-glr-example-title

.. _sphx_glr_auto_examples_snsx_plot_decileplot_script.py:


plot_decileplot_script with examples
==========================================

An example showing the :py:func:`~scikitplot.snsx.decileplot` function
used by a scikit-learn regressor.

.. GENERATED FROM PYTHON SOURCE LINES 8-12

.. code-block:: Python
   :lineno-start: 9


    # Authors: The scikit-plots developers
    # SPDX-License-Identifier: BSD-3-Clause








.. GENERATED FROM PYTHON SOURCE LINES 13-14

Import scikit-plot

.. GENERATED FROM PYTHON SOURCE LINES 14-17

.. code-block:: Python
   :lineno-start: 14

    import scikitplot.snsx as sp









.. GENERATED FROM PYTHON SOURCE LINES 18-32

.. code-block:: Python
   :lineno-start: 18

    import matplotlib.pyplot as plt
    import numpy as np; np.random.seed(0)  # reproducibility
    import pandas as pd

    from sklearn.datasets import make_classification
    from sklearn.datasets import (
        load_breast_cancer as data_2_classes,
        load_iris as data_3_classes,
        load_digits as data_10_classes,
    )
    from sklearn.linear_model import LogisticRegression
    from sklearn.model_selection import train_test_split









.. GENERATED FROM PYTHON SOURCE LINES 33-36

Load the data
X, y = data_3_classes(return_X_y=True, as_frame=False)
X, y = data_2_classes(return_X_y=True, as_frame=False)

.. GENERATED FROM PYTHON SOURCE LINES 36-43

.. code-block:: Python
   :lineno-start: 37


    # Generate a sample dataset
    X, y = make_classification(n_samples=5000, n_features=20, n_informative=15,
                              n_redundant=2, n_classes=2, n_repeated=0,
                              class_sep=1.5, flip_y=0.01, weights=[0.85, 0.15],
                              random_state=0)








.. GENERATED FROM PYTHON SOURCE LINES 44-49

.. code-block:: Python
   :lineno-start: 44

    X_train, X_val, y_train, y_val = train_test_split(
        X, y, stratify=y, test_size=0.2, random_state=0
    )
    np.unique(y)





.. rst-class:: sphx-glr-script-out

 .. code-block:: none


    array([0, 1])



.. GENERATED FROM PYTHON SOURCE LINES 50-51

Create an instance of the LogisticRegression

.. GENERATED FROM PYTHON SOURCE LINES 51-78

.. code-block:: Python
   :lineno-start: 51

    model = (
        LogisticRegression(
            # max_iter=int(1e5),
            # C=10,
            # penalty='l1',
            # solver='liblinear',
            class_weight='balanced',
            random_state=0
        )
        .fit(X_train, y_train)
    )
    # Perform predictions
    y_val_prob = model.predict_proba(X_val)
    # Create a DataFrame with predictions
    df = pd.DataFrame({
        "y_true": y_val==1,  # target class (0,1,2)
        "y_score": y_val_prob[:, 1],  # target class (0,1,2)
        # np.argmax
        "y_pred": y_val_prob[:, 1] > 0.5,  # target class (0,1,2)
        # "y_true": np.random.normal(0.5, 0.1, 100).round(),
        # "y_score": np.random.normal(0.5, 0.15, 100),
        # "hue": np.random.normal(0.5, 0.4, 100).round(),
    })
    df








.. raw:: html

    <div class="output_subarea output_html rendered_html output_result">
    <div>
    <style scoped>
        .dataframe tbody tr th:only-of-type {
            vertical-align: middle;
        }

        .dataframe tbody tr th {
            vertical-align: top;
        }

        .dataframe thead th {
            text-align: right;
        }
    </style>
    <table border="1" class="dataframe">
      <thead>
        <tr style="text-align: right;">
          <th></th>
          <th>y_true</th>
          <th>y_score</th>
          <th>y_pred</th>
        </tr>
      </thead>
      <tbody>
        <tr>
          <th>0</th>
          <td>False</td>
          <td>0.033725</td>
          <td>False</td>
        </tr>
        <tr>
          <th>1</th>
          <td>True</td>
          <td>0.860583</td>
          <td>True</td>
        </tr>
        <tr>
          <th>2</th>
          <td>False</td>
          <td>0.423101</td>
          <td>False</td>
        </tr>
        <tr>
          <th>3</th>
          <td>False</td>
          <td>0.137295</td>
          <td>False</td>
        </tr>
        <tr>
          <th>4</th>
          <td>False</td>
          <td>0.788645</td>
          <td>True</td>
        </tr>
        <tr>
          <th>...</th>
          <td>...</td>
          <td>...</td>
          <td>...</td>
        </tr>
        <tr>
          <th>995</th>
          <td>False</td>
          <td>0.228034</td>
          <td>False</td>
        </tr>
        <tr>
          <th>996</th>
          <td>False</td>
          <td>0.017187</td>
          <td>False</td>
        </tr>
        <tr>
          <th>997</th>
          <td>True</td>
          <td>0.987892</td>
          <td>True</td>
        </tr>
        <tr>
          <th>998</th>
          <td>False</td>
          <td>0.931136</td>
          <td>True</td>
        </tr>
        <tr>
          <th>999</th>
          <td>False</td>
          <td>0.128248</td>
          <td>False</td>
        </tr>
      </tbody>
    </table>
    <p>1000 rows × 3 columns</p>
    </div>
    </div>
    <br />
    <br />

.. GENERATED FROM PYTHON SOURCE LINES 79-95

.. code-block:: Python
   :lineno-start: 79

    p = sp.decileplot(
        df,
        x="y_true",
        y="y_score",
        kind="df",
        n_deciles=10,
        digits=4,
        verbose=True,
    )
    p.T
    # p.columns.tolist()
    # p[["decile", "cnt_resp", "cnt_resp_wiz", "cum_resp_pct", "cum_resp_wiz_pct"]]
    # p.iloc[:, range(9, 23)]
    # p.iloc[:, [11, 12, 12, 14]]






.. rst-class:: sphx-glr-script-out

 .. code-block:: none

    {
      "decile": "Meaning: Ranked group based on predicted probabilities (1 = highest probability). Critical: Ensure data is sorted descending by model score; top deciles should capture the majority of positives. Formula: Assign samples to k quantiles (e.g., 10 deciles) based on model score.",
      "prob_min": "Meaning: Minimum predicted probability within the decile. Critical: Indicates model calibration; values too close to prob_max suggest poor separation. Formula: min(score in decile).",
      "prob_max": "Meaning: Maximum predicted probability within the decile. Critical: Checks separation; overlap with lower deciles indicates poor discrimination. Formula: max(score in decile).",
      "prob_avg": "Meaning: Average predicted probability within the decile. Critical: Useful for calibration checks; should decrease monotonically across deciles. Formula: mean(score in decile).",
      "cnt_resp_true": "Meaning: Actual positives/responders in the decile. Critical: Should never exceed cnt_resp_wiz_true; flat counts across deciles indicate a weak or non-discriminative model. Formula: sum(y_true = 1 in decile).",
      "cnt_resp_false": "Meaning: Actual negatives/non-responders in the decile. Critical: Used in KS/statistical calculations; too many negatives in top deciles is a warning. Formula: cnt_resp_total - cnt_resp_true.",
      "cnt_resp_total": "Meaning: Total samples in the decile (positives + negatives). Critical: Denominator for rate_resp and cumulative % calculations; decile imbalance can distort lift/gain. Formula: count(samples in decile).",
      "cnt_resp_rndm_true": "Meaning: Expected positives in the decile under a random model. Critical: Baseline for lift/gain comparison; fatal if model barely exceeds random. Formula: total_positives / n_deciles.",
      "cnt_resp_wiz_true": "Meaning: Ideal/maximum possible positives if the model were perfect. Critical: Must always be ≥ cnt_resp_true; NaN or extremely low values indicate data issues. Formula: allocate top positives directly to highest scoring deciles.",
      "rate_resp": "Meaning: Decile-level response rate (alias: decile_wise_response, decile_wise_gain). Critical: Measures decile quality; early deciles should outperform later ones. Formula: rate_resp = cnt_resp_true / cnt_resp_total.",
      "overall_rate": "Meaning: Overall response rate across the dataset; serves as the baseline probability of a positive. Critical: Used as the denominator in decile-wise lift; essential to assess improvement vs random. Formula: overall_rate = sum(cnt_resp_true) / sum(cnt_resp_total) (fraction or %).",
      "cum_resp_true": "Meaning: Cumulative number of positives captured up to the current decile (alias: cumulative_gain). Critical: Should increase monotonically; maximum = total responders. Flat curve indicates weak model. Formula: Σ cnt_resp_true (≤ current decile).",
      "cum_resp_true_pct": "Meaning: Cumulative % of positives captured = cum_resp_true / total_responders * 100. Critical: Used for lift/gain curves; should always be ≥ model baseline. Formula: cum_resp_true / total_responders * 100.",
      "cum_resp_false": "Meaning: Cumulative number of negatives captured up to the current decile. Critical: Used for KS/statistical calculations; dominance in early deciles is undesirable. Formula: Σ cnt_resp_false (≤ current decile).",
      "cum_resp_false_pct": "Meaning: Cumulative % of negatives captured = cum_resp_false / total_nonresponders * 100. Critical: Should differ from cum_resp_true_pct; nearly equal curves indicate model failure. Formula: cum_resp_false / total_nonresponders * 100.",
      "cum_resp_total": "Meaning: Cumulative total samples up to the current decile. Critical: Tracks population coverage for lift/gain charts. Formula: Σ cnt_resp_total (≤ current decile).",
      "cum_resp_total_pct": "Meaning: Cumulative % of total population covered. Critical: X-axis for lift/gain curves; check decile balance. Formula: cum_resp_total / total_samples * 100.",
      "cum_resp_rndm_true": "Meaning: Cumulative expected positives if randomly assigned. Critical: Baseline for cumulative lift; fatal if model ≈ random curve. Formula: Σ cnt_resp_rndm_true (≤ current decile).",
      "cum_resp_rndm_true_pct": "Meaning: Cumulative % of expected positives under random = cum_resp_rndm_true / total_responders * 100. Critical: Baseline curve is linear from (0,0) to (100,100); model curve must exceed this. Formula: cum_resp_rndm_true / total_responders * 100.",
      "cum_resp_wiz_true": "Meaning: Cumulative ideal/maximum possible positives. Critical: Must always be ≥ model values; never NaN. Formula: Σ cnt_resp_wiz_true (≤ current decile).",
      "cum_resp_wiz_true_pct": "Meaning: % cumulative ideal positives = cum_resp_wiz_true / total_responders * 100. Critical: Wizard benchmark for lift/gain curves; gaps indicate model weakness. Formula: cum_resp_wiz_true / total_responders * 100.",
      "cumulative_lift": "Meaning: Empirical discriminative power; shows cumulative improvement vs random. Critical: Always cumulative; should exceed 1 (or ≥2 in top decile). Formula: cumulative_lift = cum_resp_true_pct / cum_resp_total_pct.",
      "decile_wise_lift": "Meaning: Improvement factor for individual deciles; shows how much better each decile performs vs random. Critical: Fatal if <1. Early deciles should show highest lift. Formula: decile_wise_lift = cnt_resp_true / cnt_resp_rndm_true.",
      "KS": "Meaning: Peak discriminative power (scalar) extracted from cumulative gain curves; maximum distance between cumulative distributions of positives and negatives. Range: 0-1 (fraction) or 0-100 (percent). Interpretation: - <0.2 → Poor discrimination - 0.2-0.4 → Fair - 0.4-0.6 → Good - ≥0.6 → Excellent - ≥0.7 → Suspiciously high (possible overfitting or leakage). Critical: Report across train/validation/test; ensure top deciles dominate appropriately. Formula: KS = max(cum_resp_true_pct - cum_resp_false_pct) (sorted descending by model score)."
    }


.. raw:: html

    <div class="output_subarea output_html rendered_html output_result">
    <div>
    <style scoped>
        .dataframe tbody tr th:only-of-type {
            vertical-align: middle;
        }

        .dataframe tbody tr th {
            vertical-align: top;
        }

        .dataframe thead th {
            text-align: right;
        }
    </style>
    <table border="1" class="dataframe">
      <thead>
        <tr style="text-align: right;">
          <th></th>
          <th>0</th>
          <th>1</th>
          <th>2</th>
          <th>3</th>
          <th>4</th>
          <th>5</th>
          <th>6</th>
          <th>7</th>
          <th>8</th>
          <th>9</th>
        </tr>
      </thead>
      <tbody>
        <tr>
          <th>decile</th>
          <td>1.0000</td>
          <td>2.0000</td>
          <td>3.0000</td>
          <td>4.0000</td>
          <td>5.0000</td>
          <td>6.0000</td>
          <td>7.0000</td>
          <td>8.0000</td>
          <td>9.0000</td>
          <td>10.0000</td>
        </tr>
        <tr>
          <th>prob_min</th>
          <td>0.8478</td>
          <td>0.5792</td>
          <td>0.3837</td>
          <td>0.2739</td>
          <td>0.2044</td>
          <td>0.1410</td>
          <td>0.0984</td>
          <td>0.0605</td>
          <td>0.0344</td>
          <td>0.0014</td>
        </tr>
        <tr>
          <th>prob_max</th>
          <td>0.9998</td>
          <td>0.8416</td>
          <td>0.5774</td>
          <td>0.3827</td>
          <td>0.2728</td>
          <td>0.2040</td>
          <td>0.1409</td>
          <td>0.0978</td>
          <td>0.0605</td>
          <td>0.0343</td>
        </tr>
        <tr>
          <th>prob_avg</th>
          <td>0.9475</td>
          <td>0.7061</td>
          <td>0.4757</td>
          <td>0.3250</td>
          <td>0.2380</td>
          <td>0.1702</td>
          <td>0.1177</td>
          <td>0.0774</td>
          <td>0.0472</td>
          <td>0.0183</td>
        </tr>
        <tr>
          <th>cnt_resp_true</th>
          <td>82.0000</td>
          <td>30.0000</td>
          <td>16.0000</td>
          <td>5.0000</td>
          <td>5.0000</td>
          <td>6.0000</td>
          <td>3.0000</td>
          <td>4.0000</td>
          <td>1.0000</td>
          <td>1.0000</td>
        </tr>
        <tr>
          <th>cnt_resp_false</th>
          <td>18.0000</td>
          <td>70.0000</td>
          <td>84.0000</td>
          <td>95.0000</td>
          <td>95.0000</td>
          <td>94.0000</td>
          <td>97.0000</td>
          <td>96.0000</td>
          <td>99.0000</td>
          <td>99.0000</td>
        </tr>
        <tr>
          <th>cnt_resp_total</th>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cnt_resp_rndm_true</th>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
        </tr>
        <tr>
          <th>cnt_resp_wiz_true</th>
          <td>100.0000</td>
          <td>53.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
        </tr>
        <tr>
          <th>rate_resp</th>
          <td>82.0000</td>
          <td>30.0000</td>
          <td>16.0000</td>
          <td>5.0000</td>
          <td>5.0000</td>
          <td>6.0000</td>
          <td>3.0000</td>
          <td>4.0000</td>
          <td>1.0000</td>
          <td>1.0000</td>
        </tr>
        <tr>
          <th>rate_resp_wiz</th>
          <td>100.0000</td>
          <td>53.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
        </tr>
        <tr>
          <th>overall_rate</th>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
          <td>15.3000</td>
        </tr>
        <tr>
          <th>cum_resp_true</th>
          <td>82.0000</td>
          <td>112.0000</td>
          <td>128.0000</td>
          <td>133.0000</td>
          <td>138.0000</td>
          <td>144.0000</td>
          <td>147.0000</td>
          <td>151.0000</td>
          <td>152.0000</td>
          <td>153.0000</td>
        </tr>
        <tr>
          <th>cum_resp_true_pct</th>
          <td>53.5948</td>
          <td>73.2026</td>
          <td>83.6601</td>
          <td>86.9281</td>
          <td>90.1961</td>
          <td>94.1176</td>
          <td>96.0784</td>
          <td>98.6928</td>
          <td>99.3464</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cum_resp_false</th>
          <td>18.0000</td>
          <td>88.0000</td>
          <td>172.0000</td>
          <td>267.0000</td>
          <td>362.0000</td>
          <td>456.0000</td>
          <td>553.0000</td>
          <td>649.0000</td>
          <td>748.0000</td>
          <td>847.0000</td>
        </tr>
        <tr>
          <th>cum_resp_false_pct</th>
          <td>2.1251</td>
          <td>10.3896</td>
          <td>20.3070</td>
          <td>31.5230</td>
          <td>42.7391</td>
          <td>53.8371</td>
          <td>65.2893</td>
          <td>76.6234</td>
          <td>88.3117</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cum_resp_total</th>
          <td>100.0000</td>
          <td>200.0000</td>
          <td>300.0000</td>
          <td>400.0000</td>
          <td>500.0000</td>
          <td>600.0000</td>
          <td>700.0000</td>
          <td>800.0000</td>
          <td>900.0000</td>
          <td>1000.0000</td>
        </tr>
        <tr>
          <th>cum_resp_total_pct</th>
          <td>10.0000</td>
          <td>20.0000</td>
          <td>30.0000</td>
          <td>40.0000</td>
          <td>50.0000</td>
          <td>60.0000</td>
          <td>70.0000</td>
          <td>80.0000</td>
          <td>90.0000</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cum_resp_rndm_true</th>
          <td>15.3000</td>
          <td>30.6000</td>
          <td>45.9000</td>
          <td>61.2000</td>
          <td>76.5000</td>
          <td>91.8000</td>
          <td>107.1000</td>
          <td>122.4000</td>
          <td>137.7000</td>
          <td>153.0000</td>
        </tr>
        <tr>
          <th>cum_resp_rndm_true_pct</th>
          <td>10.0000</td>
          <td>20.0000</td>
          <td>30.0000</td>
          <td>40.0000</td>
          <td>50.0000</td>
          <td>60.0000</td>
          <td>70.0000</td>
          <td>80.0000</td>
          <td>90.0000</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cum_resp_wiz_true</th>
          <td>100.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
          <td>153.0000</td>
        </tr>
        <tr>
          <th>cum_resp_wiz_true_pct</th>
          <td>65.3595</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
          <td>100.0000</td>
        </tr>
        <tr>
          <th>cumulative_lift</th>
          <td>5.3595</td>
          <td>3.6601</td>
          <td>2.7887</td>
          <td>2.1732</td>
          <td>1.8039</td>
          <td>1.5686</td>
          <td>1.3725</td>
          <td>1.2337</td>
          <td>1.1038</td>
          <td>1.0000</td>
        </tr>
        <tr>
          <th>cumulative_lift_wiz</th>
          <td>6.5359</td>
          <td>5.0000</td>
          <td>3.3333</td>
          <td>2.5000</td>
          <td>2.0000</td>
          <td>1.6667</td>
          <td>1.4286</td>
          <td>1.2500</td>
          <td>1.1111</td>
          <td>1.0000</td>
        </tr>
        <tr>
          <th>decile_wise_lift</th>
          <td>5.3595</td>
          <td>1.9608</td>
          <td>1.0458</td>
          <td>0.3268</td>
          <td>0.3268</td>
          <td>0.3922</td>
          <td>0.1961</td>
          <td>0.2614</td>
          <td>0.0654</td>
          <td>0.0654</td>
        </tr>
        <tr>
          <th>decile_wise_lift_wiz</th>
          <td>6.5359</td>
          <td>3.4641</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
          <td>0.0000</td>
        </tr>
        <tr>
          <th>KS</th>
          <td>51.4696</td>
          <td>62.8130</td>
          <td>63.3532</td>
          <td>55.4051</td>
          <td>47.4570</td>
          <td>40.2806</td>
          <td>30.7892</td>
          <td>22.0694</td>
          <td>11.0347</td>
          <td>0.0000</td>
        </tr>
      </tbody>
    </table>
    </div>
    </div>
    <br />
    <br />

.. GENERATED FROM PYTHON SOURCE LINES 96-98

.. code-block:: Python
   :lineno-start: 96

    p = sp.decileplot(df, x="y_true", y="y_score", kind="cumulative_lift", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_001.png
   :alt: Cumulative Lift Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_001.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 99-101

.. code-block:: Python
   :lineno-start: 99

    p = sp.decileplot(df, x="y_true", y="y_score", kind="decile_wise_lift", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_002.png
   :alt: Decile-wise Lift Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_002.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 102-104

.. code-block:: Python
   :lineno-start: 102

    p = sp.decileplot(df, x="y_true", y="y_score", kind="cumulative_gain", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_003.png
   :alt: Cumulative Gain Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_003.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 105-107

.. code-block:: Python
   :lineno-start: 105

    p = sp.decileplot(df, x="y_true", y="y_score", kind="cumulative_response", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_004.png
   :alt: Cumulative Response Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_004.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 108-110

.. code-block:: Python
   :lineno-start: 108

    p = sp.decileplot(df, x="y_true", y="y_score", kind="decile_wise_gain", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_005.png
   :alt: Decile-wise Gain/Response Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_005.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 111-113

.. code-block:: Python
   :lineno-start: 111

    p = sp.decileplot(df, x="y_true", y="y_score", kind="ks_statistic", n_deciles=10, annot=True)




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_006.png
   :alt: Kolmogorov-Smirnov (KS) Statistic Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_006.png
   :class: sphx-glr-single-img





.. GENERATED FROM PYTHON SOURCE LINES 114-115

fig, ax = plt.subplots(figsize=(10, 10))

.. GENERATED FROM PYTHON SOURCE LINES 115-126

.. code-block:: Python
   :lineno-start: 115

    p = sp.decileplot(
        df,
        x="y_true",
        y="y_score",
        kind="report",
        n_deciles=10,
        digits=4,
        annot=True,
        verbose=True,
    )




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_007.png
   :alt: Cumulative Lift Curve, Decile-wise Lift Curve, Cumulative Gain Curve, Kolmogorov-Smirnov (KS) Statistic Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_007.png
   :class: sphx-glr-single-img


.. rst-class:: sphx-glr-script-out

 .. code-block:: none

    {
      "decile": "Meaning: Ranked group based on predicted probabilities (1 = highest probability). Critical: Ensure data is sorted descending by model score; top deciles should capture the majority of positives. Formula: Assign samples to k quantiles (e.g., 10 deciles) based on model score.",
      "prob_min": "Meaning: Minimum predicted probability within the decile. Critical: Indicates model calibration; values too close to prob_max suggest poor separation. Formula: min(score in decile).",
      "prob_max": "Meaning: Maximum predicted probability within the decile. Critical: Checks separation; overlap with lower deciles indicates poor discrimination. Formula: max(score in decile).",
      "prob_avg": "Meaning: Average predicted probability within the decile. Critical: Useful for calibration checks; should decrease monotonically across deciles. Formula: mean(score in decile).",
      "cnt_resp_true": "Meaning: Actual positives/responders in the decile. Critical: Should never exceed cnt_resp_wiz_true; flat counts across deciles indicate a weak or non-discriminative model. Formula: sum(y_true = 1 in decile).",
      "cnt_resp_false": "Meaning: Actual negatives/non-responders in the decile. Critical: Used in KS/statistical calculations; too many negatives in top deciles is a warning. Formula: cnt_resp_total - cnt_resp_true.",
      "cnt_resp_total": "Meaning: Total samples in the decile (positives + negatives). Critical: Denominator for rate_resp and cumulative % calculations; decile imbalance can distort lift/gain. Formula: count(samples in decile).",
      "cnt_resp_rndm_true": "Meaning: Expected positives in the decile under a random model. Critical: Baseline for lift/gain comparison; fatal if model barely exceeds random. Formula: total_positives / n_deciles.",
      "cnt_resp_wiz_true": "Meaning: Ideal/maximum possible positives if the model were perfect. Critical: Must always be ≥ cnt_resp_true; NaN or extremely low values indicate data issues. Formula: allocate top positives directly to highest scoring deciles.",
      "rate_resp": "Meaning: Decile-level response rate (alias: decile_wise_response, decile_wise_gain). Critical: Measures decile quality; early deciles should outperform later ones. Formula: rate_resp = cnt_resp_true / cnt_resp_total.",
      "overall_rate": "Meaning: Overall response rate across the dataset; serves as the baseline probability of a positive. Critical: Used as the denominator in decile-wise lift; essential to assess improvement vs random. Formula: overall_rate = sum(cnt_resp_true) / sum(cnt_resp_total) (fraction or %).",
      "cum_resp_true": "Meaning: Cumulative number of positives captured up to the current decile (alias: cumulative_gain). Critical: Should increase monotonically; maximum = total responders. Flat curve indicates weak model. Formula: Σ cnt_resp_true (≤ current decile).",
      "cum_resp_true_pct": "Meaning: Cumulative % of positives captured = cum_resp_true / total_responders * 100. Critical: Used for lift/gain curves; should always be ≥ model baseline. Formula: cum_resp_true / total_responders * 100.",
      "cum_resp_false": "Meaning: Cumulative number of negatives captured up to the current decile. Critical: Used for KS/statistical calculations; dominance in early deciles is undesirable. Formula: Σ cnt_resp_false (≤ current decile).",
      "cum_resp_false_pct": "Meaning: Cumulative % of negatives captured = cum_resp_false / total_nonresponders * 100. Critical: Should differ from cum_resp_true_pct; nearly equal curves indicate model failure. Formula: cum_resp_false / total_nonresponders * 100.",
      "cum_resp_total": "Meaning: Cumulative total samples up to the current decile. Critical: Tracks population coverage for lift/gain charts. Formula: Σ cnt_resp_total (≤ current decile).",
      "cum_resp_total_pct": "Meaning: Cumulative % of total population covered. Critical: X-axis for lift/gain curves; check decile balance. Formula: cum_resp_total / total_samples * 100.",
      "cum_resp_rndm_true": "Meaning: Cumulative expected positives if randomly assigned. Critical: Baseline for cumulative lift; fatal if model ≈ random curve. Formula: Σ cnt_resp_rndm_true (≤ current decile).",
      "cum_resp_rndm_true_pct": "Meaning: Cumulative % of expected positives under random = cum_resp_rndm_true / total_responders * 100. Critical: Baseline curve is linear from (0,0) to (100,100); model curve must exceed this. Formula: cum_resp_rndm_true / total_responders * 100.",
      "cum_resp_wiz_true": "Meaning: Cumulative ideal/maximum possible positives. Critical: Must always be ≥ model values; never NaN. Formula: Σ cnt_resp_wiz_true (≤ current decile).",
      "cum_resp_wiz_true_pct": "Meaning: % cumulative ideal positives = cum_resp_wiz_true / total_responders * 100. Critical: Wizard benchmark for lift/gain curves; gaps indicate model weakness. Formula: cum_resp_wiz_true / total_responders * 100.",
      "cumulative_lift": "Meaning: Empirical discriminative power; shows cumulative improvement vs random. Critical: Always cumulative; should exceed 1 (or ≥2 in top decile). Formula: cumulative_lift = cum_resp_true_pct / cum_resp_total_pct.",
      "decile_wise_lift": "Meaning: Improvement factor for individual deciles; shows how much better each decile performs vs random. Critical: Fatal if <1. Early deciles should show highest lift. Formula: decile_wise_lift = cnt_resp_true / cnt_resp_rndm_true.",
      "KS": "Meaning: Peak discriminative power (scalar) extracted from cumulative gain curves; maximum distance between cumulative distributions of positives and negatives. Range: 0-1 (fraction) or 0-100 (percent). Interpretation: - <0.2 → Poor discrimination - 0.2-0.4 → Fair - 0.4-0.6 → Good - ≥0.6 → Excellent - ≥0.7 → Suspiciously high (possible overfitting or leakage). Critical: Report across train/validation/test; ensure top deciles dominate appropriately. Formula: KS = max(cum_resp_true_pct - cum_resp_false_pct) (sorted descending by model score)."
    }
       decile  prob_min  prob_max  ...  decile_wise_lift  decile_wise_lift_wiz       KS
    0       1    0.8478    0.9998  ...            5.3595                6.5359  51.4696
    1       2    0.5792    0.8416  ...            1.9608                3.4641  62.8130
    2       3    0.3837    0.5774  ...            1.0458                0.0000  63.3532
    3       4    0.2739    0.3827  ...            0.3268                0.0000  55.4051
    4       5    0.2044    0.2728  ...            0.3268                0.0000  47.4570
    5       6    0.1410    0.2040  ...            0.3922                0.0000  40.2806
    6       7    0.0984    0.1409  ...            0.1961                0.0000  30.7892
    7       8    0.0605    0.0978  ...            0.2614                0.0000  22.0694
    8       9    0.0344    0.0605  ...            0.0654                0.0000  11.0347
    9      10    0.0014    0.0343  ...            0.0654                0.0000   0.0000

    [10 rows x 27 columns]




.. GENERATED FROM PYTHON SOURCE LINES 127-128

fig, ax = plt.subplots(figsize=(10, 10))

.. GENERATED FROM PYTHON SOURCE LINES 128-138

.. code-block:: Python
   :lineno-start: 128

    p = sp.decileplot(
        df,
        x="y_true",
        y="y_score",
        kind="report",
        n_deciles=10,
        digits=6,
        fmt='.4g'
    )




.. image-sg:: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_008.png
   :alt: Cumulative Lift Curve, Decile-wise Lift Curve, Cumulative Gain Curve, Kolmogorov-Smirnov (KS) Statistic Curve
   :srcset: /auto_examples/snsx/images/sphx_glr_plot_decileplot_script_008.png
   :class: sphx-glr-single-img


.. rst-class:: sphx-glr-script-out

 .. code-block:: none

       decile  prob_min  ...  decile_wise_lift_wiz         KS
    0       1  0.847810  ...              6.535948  51.469624
    1       2  0.579150  ...              3.464052  62.813004
    2       3  0.383729  ...              0.000000  63.353165
    3       4  0.273915  ...              0.000000  55.405082
    4       5  0.204425  ...              0.000000  47.456999
    5       6  0.141044  ...              0.000000  40.280575
    6       7  0.098419  ...              0.000000  30.789175
    7       8  0.060545  ...              0.000000  22.069434
    8       9  0.034429  ...              0.000000  11.034717
    9      10  0.001371  ...              0.000000   0.000000

    [10 rows x 27 columns]




.. GENERATED FROM PYTHON SOURCE LINES 139-147

.. tags::

   model-type: classification
   model-workflow: model evaluation
   plot-type: line
   plot-type: decile
   level: beginner
   purpose: showcase


.. rst-class:: sphx-glr-timing

   **Total running time of the script:** (0 minutes 1.954 seconds)


.. _sphx_glr_download_auto_examples_snsx_plot_decileplot_script.py:

.. only:: html

  .. container:: sphx-glr-footer sphx-glr-footer-example

    .. container:: binder-badge

      .. image:: images/binder_badge_logo.svg
        :target: https://mybinder.org/v2/gh/scikit-plots/scikit-plots/maintenance/0.4.X?urlpath=lab/tree/notebooks/auto_examples/snsx/plot_decileplot_script.ipynb
        :alt: Launch binder
        :width: 150 px

    .. container:: lite-badge

      .. image:: images/jupyterlite_badge_logo.svg
        :target: ../../lite/lab/index.html?path=auto_examples/snsx/plot_decileplot_script.ipynb
        :alt: Launch JupyterLite
        :width: 150 px

    .. container:: sphx-glr-download sphx-glr-download-jupyter

      :download:`Download Jupyter notebook: plot_decileplot_script.ipynb <plot_decileplot_script.ipynb>`

    .. container:: sphx-glr-download sphx-glr-download-python

      :download:`Download Python source code: plot_decileplot_script.py <plot_decileplot_script.py>`

    .. container:: sphx-glr-download sphx-glr-download-zip

      :download:`Download zipped: plot_decileplot_script.zip <plot_decileplot_script.zip>`


.. include:: plot_decileplot_script.recommendations


.. only:: html

 .. rst-class:: sphx-glr-signature

    `Gallery generated by Sphinx-Gallery <https://sphinx-gallery.github.io>`_