Probability calibration in ML Python - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to import the calibration module from scikit-learn.

ML Python

from sklearn.calibration import [1]

Drag options to blanks, or click blank then click option'

ACalibratedClassifierCV

Btrain_test_split

CStandardScaler

DLogisticRegression

Attempts:

3 left

💡 Hint

Common Mistakes

Importing unrelated classes like train_test_split or StandardScaler.

Confusing model classes with calibration classes.

The CalibratedClassifierCV class is used for probability calibration in scikit-learn.

2fill in blank

medium

Complete the code to create a calibrated classifier using sigmoid method.

ML Python

calibrated_clf = CalibratedClassifierCV(base_estimator=clf, method=[1], cv='prefit')

Drag options to blanks, or click blank then click option'

Apolynomial

Bisotonic

Clinear

Dsigmoid

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'isotonic' when sigmoid is required.

Choosing methods not supported by CalibratedClassifierCV.

The sigmoid method applies Platt scaling for probability calibration.

3fill in blank

hard

Fix the error in the code to correctly fit the calibrated classifier.

ML Python

calibrated_clf.[1](X_calib, y_calib)

Drag options to blanks, or click blank then click option'

Apredict

Bfit

Cpredict_proba

Dscore

Attempts:

3 left

💡 Hint

Common Mistakes

Calling predict or predict_proba before fitting the model.

Using score method which only evaluates performance.

The fit method trains the calibrated classifier on calibration data.

4fill in blank

hard

Fill both blanks to create a dictionary of calibrated probabilities for each class.

ML Python

calibrated_probs = {cls: calibrated_clf.predict_proba(X)[:, [1]] for [2] in calibrated_clf.classes_}

Drag options to blanks, or click blank then click option'

A0

B1

Ccls

Dclass_label

Attempts:

3 left

💡 Hint

Common Mistakes

Using a fixed index like 0 or 1 instead of the loop variable.

Using a variable name not defined in the comprehension.

cls selects the probability for the corresponding class, and class_label is the variable iterating over classes.

5fill in blank

hard

Fill all three blanks to compute Brier score loss for calibrated predictions.

ML Python

from sklearn.metrics import [1]
brier_score = [2](y_true, calibrated_clf.[3](X_test)[:, 1])

Drag options to blanks, or click blank then click option'

Aaccuracy_score

Bbrier_score_loss

Cpredict_proba

Dpredict

Attempts:

3 left

💡 Hint

Common Mistakes

Using accuracy_score which measures classification accuracy, not probability calibration.

Using predict instead of predict_proba which returns class labels.

The brier_score_loss function measures the accuracy of probabilistic predictions, which are obtained using predict_proba.