Objective 3.3 Carry Out Feature Engineering CERTIFIED-DATA-SCIENCE-PRACTITIONER Practice Questions (Page 6)

Questions 26–30

1. 26foundation · easy

   A data scientist is building a linear regression model and wants to encode a categorical feature with 5 categories. To avoid multicollinearity, how many binary columns should they create?

   • A1
   • B4
   • C5
   • D3
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   A dataset contains a numeric feature where all values are nearly constant, with very little variation. The data scientist wants to remove such features because they carry little information. Which technique should they use?

   • ATarget encoding
   • BOne-hot encoding
   • CPCA
   • DLow-variance filter
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   A data scientist wants to encode a high-cardinality categorical feature by replacing each category with the average of the target variable for that category. They are concerned about overfitting, especially for categories with few samples. Which technique should they use?

   • ATarget encoding with smoothing
   • BLabel encoding
   • COne-hot encoding
   • DBinary encoding
   Reveal Solution Discussion Select an answer first
5. 29foundation · easy

   A dataset contains a categorical feature with thousands of unique values. The data scientist wants to encode it into a fixed number of columns without creating a separate column for each category. Which encoding method is most suitable?

   • ALabel encoding
   • BHash encoding
   • COne-hot encoding
   • DTarget encoding
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A data scientist is preparing a dataset for a logistic regression model to predict customer churn. The dataset includes a 'Region' column with values 'North', 'South', 'East', and 'West'. The model will be deployed in production and must be interpretable. What encoding approach should be used?

   • ADummy encoding, creating three binary columns
   • BOne-hot encoding, creating four binary columns
   • CTarget encoding, using the mean churn rate per region
   • DLabel encoding, assigning values 0, 1, 2, 3
   Reveal Solution Discussion Select an answer first