Manage Data Transformation to Conduct Data Preparation PMI-CPMAI Practice Questions (Page 6)

Questions 26–30

1. 26application · medium

   A data engineering team is building a preprocessing pipeline for a customer segmentation model. The raw data contains missing values in the 'income' column, duplicate records, and some extreme outliers in 'age'. The team wants to ensure the pipeline systematically detects and corrects these issues before modeling. Which approach should they implement?

   • ACreate a data cleaning workflow that handles missing values, duplicates, and outliers in a defined sequence
   • BIgnore missing values and outliers and train the model directly
   • CApply a single imputation method to all missing values without checking for duplicates
   • DManually clean the data in a spreadsheet before loading it into the pipeline
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   What is the purpose of a data preprocessing pipeline?

   • ATo deploy the trained model to production
   • BTo transform raw data into a clean, usable format for model training
   • CTo collect raw data from various sources
   • DTo evaluate model performance on test data
   Reveal Solution Discussion Select an answer first
4. 28foundation · easy

   What is the purpose of data augmentation?

   • ATo generate additional training samples to improve model generalization
   • BTo clean the data by removing duplicates
   • CTo reduce the number of features in the dataset
   • DTo scale features to a comparable range
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A machine learning team is developing a model that will be retrained monthly. They need to ensure that the data transformation steps (e.g., imputation, scaling) produce the same results every time the pipeline is run, so that model performance is comparable across retraining cycles. Which practice should they adopt?

   • AUse a different random seed each time the pipeline runs
   • BOnly version the code, not the data
   • CSet a fixed random seed for all random operations and version the data and code
   • DManually adjust the transformation parameters each month
   Reveal Solution Discussion Select an answer first
6. 30foundation · easy

   What is the purpose of feature engineering?

   • ATo create informative features from raw data to improve model performance
   • BTo scale features to a common range
   • CTo clean the data by removing missing values
   • DTo reduce the number of features by selecting a subset
   Reveal Solution Discussion Select an answer first