Boosting concept in ML Python - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a simple AdaBoost classifier using scikit-learn.

ML Python

from sklearn.ensemble import AdaBoostClassifier
from sklearn.tree import DecisionTreeClassifier

model = AdaBoostClassifier(base_estimator=[1], n_estimators=50, random_state=42)
model.fit(X_train, y_train)

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

AKNeighborsClassifier()

BRandomForestClassifier()

CDecisionTreeClassifier(max_depth=1)

DSVC()

Attempts:

3 left

💡 Hint

Common Mistakes

Using complex models like RandomForest or SVC as base estimators.

Not specifying max_depth=1 for the decision tree.

AdaBoost commonly uses a simple decision tree with max_depth=1 as its base estimator, called a stump.

2fill in blank

medium

Complete the code to calculate the weighted error of a weak learner in boosting.

ML Python

weighted_error = sum(sample_weights * (predictions != y_true)) / [1]

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

Asum(sample_weights)

Blen(predictions)

Csum(y_true)

Dlen(sample_weights)

Attempts:

3 left

💡 Hint

Common Mistakes

Dividing by the number of samples instead of sum of weights.

Using sum(y_true) which is unrelated.

The weighted error divides the weighted sum of wrong predictions by the total sum of weights.

3fill in blank

hard

Fix the error in updating sample weights after a boosting iteration.

ML Python

sample_weights = sample_weights * np.exp([1] * (predictions != y_true))
sample_weights /= sample_weights.sum()

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

A1

B-alpha

Calpha

D-1

Attempts:

3 left

💡 Hint

Common Mistakes

Using positive alpha which decreases weights for errors.

Using constants 1 or -1 which ignore alpha.

Weights increase for misclassified samples, so the exponent must be negative alpha times the error indicator.

4fill in blank

hard

Fill both blanks to compute the alpha (learner weight) in AdaBoost.

ML Python

alpha = 0.5 * np.log((1 - [1]) / [2])

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

Aweighted_error

B1 - weighted_error

D1 + weighted_error

Attempts:

3 left

💡 Hint

Common Mistakes

Swapping numerator and denominator.

Using incorrect expressions like 1 + weighted_error.

Alpha is calculated as 0.5 * log((1 - error) / error) where error is the weighted error.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that maps each weak learner to its alpha weight if alpha is positive.

ML Python

learner_weights = [1]: [2] for learner, [3] in zip(learners, alphas) if [2] > 0}

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

Alearner

Balpha

Dweight

Attempts:

3 left

💡 Hint

Common Mistakes

Using wrong variable names like 'weight' or inconsistent names.

Not filtering by positive alpha.

The dictionary maps each learner to its alpha weight, filtering only positive alphas.