Virtual Labs

Dimensionality Reduction: Principal Component Analysis (PCA)

In the experiment, why is the dataset standardized before applying Principal Component Analysis?

a: To remove correlated features Explanation

Explanation

b: To convert categorical variables into numerical form Explanation

Explanation

c: To ensure that all features contribute equally to the computation of principal components Explanation

Explanation

d: To reduce the number of samples Explanation

Explanation

Which observation is made when the shape of the dataset is displayed before applying PCA?

a: The number of samples is reduced Explanation

Explanation

b: The target labels are removed Explanation

Explanation

c: The number of features decreases after PCA transformation Explanation

Explanation

d: The dataset becomes unbalanced Explanation

Explanation

Why is feature correlation an important factor when deciding to apply PCA?

a: PCA works only on uncorrelated features Explanation

Explanation

b: Highly correlated features contain redundant information Explanation

Explanation

c: Correlation improves classification accuracy Explanation

Explanation

d: Correlation removes noise from data Explanation

Explanation

Which mathematical property allows PCA to transform original features into independent components?

a: Feature normalization Explanation

Explanation

b: Eigen decomposition of the covariance matrix Explanation

Explanation

c: Gradient descent optimization Explanation

Explanation

d: Probability density estimation Explanation

Explanation

What is the implication of choosing a 95% explained variance threshold in PCA?

a: Exactly 95% of features are retained Explanation

Explanation

b: Exactly 95 principal components are created Explanation

Explanation

c: Enough components are selected to preserve 95% of data variability Explanation

Explanation

d: The classifier will achieve 95% accuracy Explanation

Explanation

Why is PCA considered an unsupervised dimensionality reduction technique?

a: Because it removes noisy labels Explanation

Explanation

b: Because it improves supervised learning models Explanation

Explanation

c: Because it does not use class labels while learning transformations Explanation

Explanation

d: Because it balances the dataset Explanation

Explanation

Why is classification accuracy evaluated both before and after PCA?

a: To confirm that PCA increases training complexity Explanation

Explanation

b: To ensure that dimensionality reduction does not degrade predictive performance Explanation

Explanation

c: To eliminate bias in dataset Explanation

Explanation

d: To increase the number of training samples Explanation

Explanation

Which statement best explains why PCA can improve computational efficiency?

a: It increases the number of training samples Explanation

Explanation

b: It simplifies the decision boundary Explanation

Explanation

c: It reduces dimensionality, lowering computation and memory cost Explanation

Explanation

d: It converts nonlinear data into linear form Explanation

Explanation

What does a low explained variance ratio for a principal component indicate?

a: The component has high predictive power Explanation

Explanation

b: The component captures little information from the data Explanation

Explanation

c: The component removes noise completely Explanation

Explanation

d: The component improves classification accuracy Explanation

Explanation

Why are PCA feature loadings important for model interpretability?

a: They remove irrelevant samples Explanation

Explanation

b: They show how original features contribute to each principal component Explanation

Explanation

c: They improve classifier accuracy Explanation

Explanation

d: They normalize the dataset Explanation

Explanation