Machine Learning Supervised Test

Using Principal Component Analysis (PCA).

Which of the following methods is NOT commonly used to address multicollinearity in logistic regression?

The model performs better on unseen data.

What happens when K is too small in the KNN algorithm?

What is the time complexity of the KNN algorithm during prediction (assuming no optimizations)?

It makes distance-based computations less meaningful

It speeds up nearest-neighbor calculations

How does the curse of dimensionality affect KNN?

Boosting focuses on misclassified samples, leading to better generalization.

Boosting increases the diversity of base models.

Boosting reduces the number of required models.

Why does boosting often outperform bagging in practice?

Which of the following algorithms is NOT a boosting technique?

It does not work with categorical variables.

It cannot be used for classification problems.

It requires a large number of features to work.

Which of the following is a key limitation of logistic regression?

It maximizes the probability of observed outcomes given the parameters.

How does Maximum Likelihood Estimation (MLE) help in logistic regression?

Which distance metric is NOT typically used in KNN?

By adjusting the learning rate of each tree.

By fitting to the residual errors of the previous predictions.

By increasing the depth of trees dynamically.

In Gradient Boosting Machines (GBM), how does each new tree correct the errors of the previous trees?

By increasing the weight of correctly classified samples.

By reducing the learning rate of each weak learner.

By increasing the weight of misclassified samples.

By adjusting the depth of decision trees dynamically.

In AdaBoost, how are weak learners updated during training?

The model oscillates and may not converge.

In a logistic regression model, what happens if we use an extremely large learning rate?

In logistic regression, why do we use the sigmoid function?

Using dimensionality reduction techniques like PCA.

Removing test data points before training.

Which technique can be used to speed up KNN for large datasets?



To reduce variance by training multiple models independently.

To increase bias by aggregating weak models.

To improve interpretability of individual models.

To make training faster by using fewer data samples.

What is the primary goal of bagging in ensemble learning?