Data Science Feature Engineering and Selection 5

Question 1

Tree-based models provide feature importance primarily based on:

Accepted Answer

How much each feature reduces impurity across splits

Answer

Alphabetical order of features

Answer

The feature's mean value

Answer

The number of missing values

Question 2

Permutation importance measures a feature's value by:

Accepted Answer

Shuffling its values and observing the drop in model performance

Answer

Removing all rows with that feature

Answer

Scaling it to unit variance

Answer

Counting its unique categories

Question 3

Which dimensionality reduction technique is unsupervised and linear?

Accepted Answer

Principal Component Analysis (PCA)

Answer

Recursive Feature Elimination

Answer

Lasso regression

Answer

Target encoding

Question 4

A key difference between PCA and feature selection is that PCA:

Accepted Answer

Creates new combined features rather than keeping original ones

Answer

Always keeps the original interpretable features

Answer

Requires the target variable

Answer

Cannot reduce dimensions

Question 5

For text data, TF-IDF weighting down-weights words that:

Accepted Answer

Appear frequently across many documents

Answer

Appear in only one document

Answer

Are numerically encoded

Answer

Have high variance

Question 6

Feature hashing (the hashing trick) is mainly used to:

Accepted Answer

Map high-cardinality categories to a fixed-size space efficiently

Answer

Encrypt the dataset

Answer

Remove missing values

Answer

Guarantee zero collisions

Question 7

When using SHAP values for feature analysis, they explain:

Accepted Answer

Each feature's contribution to an individual prediction

Answer

Only the global dataset mean

Answer

The training time per feature

Answer

The number of categories