Data Science FREE Data Science Feature Engineering and Selection Questions and Answers 2 — Questions and Answers
Question 1: Which technique creates new features by combining two or more existing categorical variables into a single feature?
- Feature crossing (Correct answer)
- Principal Component Analysis
- Normalization
- Recursive feature elimination
Correct answer: Feature crossing
Feature crossing combines categorical variables to capture interaction effects between them.
Question 2: What is the primary purpose of using Variance Inflation Factor (VIF) in feature selection?
- To detect multicollinearity among predictor variables (Correct answer)
- To measure feature importance in tree models
- To normalize skewed distributions
- To reduce the number of target classes
Correct answer: To detect multicollinearity among predictor variables
VIF quantifies how much a feature's variance is inflated due to correlation with other predictors.
Question 3: When performing target encoding on a categorical feature, what problem can arise if no regularization is applied?
- Target leakage leading to overfitting (Correct answer)
- Loss of all categorical information
- Automatic removal of rare categories
- Conversion of numeric features to categorical
Correct answer: Target leakage leading to overfitting
Without regularization, target encoding can leak target information into features, causing severe overfitting.
Question 4: Which feature selection method evaluates subsets of features by actually training a model and measuring performance?
- Wrapper method (Correct answer)
- Filter method
- Embedded method
- Transformation method
Correct answer: Wrapper method
Wrapper methods use a predictive model to score feature subsets and select the best-performing combination.
Question 5: What is the main advantage of using mutual information over Pearson correlation for feature selection?
- It captures non-linear relationships between variables (Correct answer)
- It runs faster on large datasets
- It only works with continuous features
- It requires no hyperparameter tuning
Correct answer: It captures non-linear relationships between variables
Mutual information measures any statistical dependency between variables, not just linear relationships.
Question 6: In time-series feature engineering, what does a lag feature represent?
- The value of a variable at a previous time step (Correct answer)
- The cumulative sum of all prior values
- The difference between the maximum and minimum values
- The seasonally adjusted trend component
Correct answer: The value of a variable at a previous time step
A lag feature shifts a variable's value by one or more time steps to capture temporal dependencies.
Data Science FREE Data Science Feature Engineering and Selection Questions and Answers 2
Which technique creates new features by combining two or more existing categorical variables into a single feature?