Logistic Regression

We use a classification model to predict which customers will default on their credit card debt.

Data

To learn more about the data and all of the data preparation steps, take a look at this page. Here, we simply import the prepared data:

import pandas as pd

df = pd.read_csv('https://raw.githubusercontent.com/kirenz/classification/main/_static/data/default-prepared.csv')

# preparation of label and features
y = df['default_Yes']
X = df.drop(columns = 'default_Yes')

Data split

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state = 1)

Model

from sklearn.linear_model import LogisticRegression

clf = LogisticRegression()

clf.fit(X_train, y_train)

LogisticRegression()

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

y_pred = clf.predict(X_test)

# Return the mean accuracy on the given test data and labels:
clf.score(X_test, y_test)

0.968

Confusion matrix

from sklearn.metrics import ConfusionMatrixDisplay

ConfusionMatrixDisplay.from_estimator(clf, X_test, y_test);

Classification report

from sklearn.metrics import classification_report

print(classification_report(y_test, y_pred, target_names=['0', '1']))

              precision    recall  f1-score   support

           0       0.97      0.99      0.98      2909
           1       0.43      0.18      0.25        91

    accuracy                           0.97      3000
   macro avg       0.70      0.58      0.62      3000
weighted avg       0.96      0.97      0.96      3000

ROC Curve

from sklearn.metrics import RocCurveDisplay

RocCurveDisplay.from_estimator(clf, X_test, y_test);

AUC Score

from sklearn.metrics import roc_auc_score

roc_auc_score(y_test, clf.decision_function(X_test))

0.8946921074800072

y_score = clf.predict_proba(X_test)[:, 1]
roc_auc_score(y_test, y_score)

0.8946921074800072

Change threshold

Use specific threshold

# obtain probabilities
pred_proba = clf.predict_proba(X_test)

# set threshold to 0.25

df_25 = pd.DataFrame({'y_test': y_test, 'y_pred': pred_proba[:,1] > .25})

ConfusionMatrixDisplay.from_predictions(y_test, df_25['y_pred']);

Classification report

print(classification_report(y_test, df_25['y_pred'], target_names=['0', '1']))

              precision    recall  f1-score   support

           0       0.98      0.97      0.97      2909
           1       0.29      0.41      0.34        91

    accuracy                           0.95      3000
   macro avg       0.64      0.69      0.66      3000
weighted avg       0.96      0.95      0.96      3000