Objective 2.4 Transform Numerical and Categorical Data AIP-210 Practice Questions (Page 4)

Questions 16–20

1. 16foundation · easy

   In a data transformation pipeline, why is it important to fit the transformation parameters (e.g., mean and standard deviation for standardization) only on the training data and then apply the same parameters to the test data?

   • ATo improve the accuracy of the model on the training data
   • BTo reduce the training time
   • CTo make the test data larger
   • DTo ensure the test data is transformed identically to the training data, preventing data leakage
   Reveal Solution Discussion Select an answer first
2. 17foundation · easy

   A data scientist is preparing features for a linear regression model and notices that one feature has a highly skewed distribution. Which numerical transformation is most appropriate to reduce the skewness before modeling?

   • ARobust scaling
   • BMin-max scaling
   • CStandardization (z-score)
   • DLog transformation
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A machine learning engineer is building a classifier to predict customer churn. The dataset includes a 'CustomerSegment' feature with categories: 'Basic', 'Standard', 'Premium', and 'Enterprise'. The categories have a natural ordering from least to most valuable. The model will be a decision tree, which does not assume linear relationships. Which encoding method should the engineer use for this feature?

   • AOne-hot encoding
   • BFrequency encoding
   • CLabel encoding (arbitrary integers)
   • DOrdinal encoding with the correct order
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which binning approach divides the range of a continuous feature into intervals that each contain approximately the same number of observations?

   • AEqual-width binning
   • BQuantile binning
   • CEqual-frequency binning
   • DLogarithmic binning
   Reveal Solution Discussion Select an answer first
6. 20application · medium

   A data scientist is building a regression model to predict house prices. The 'SquareFootage' feature has a few extreme values (e.g., 10,000 sq ft) that are legitimate luxury homes, not data errors. The model is a linear regression, and these outliers are heavily influencing the coefficient estimates. The team wants to keep the data points but reduce their impact. Which approach should the data scientist take?

   • AApply a log transformation to the SquareFootage feature.
   • BStandardize the feature using z-scores.
   • CCap the SquareFootage values at the 95th percentile.
   • DRemove the outlier rows from the dataset.
   Reveal Solution Discussion Select an answer first