Data Cleaning and Standardization PCAD-31 Practice Questions (Page 2)

Questions 6–10

1. 6expert · hard

   A data analyst is preparing a dataset for a clustering algorithm. The dataset has two features: 'salary' (with a few extreme high values) and 'years_of_experience' (with no outliers). The analyst wants to use k-means clustering, which is sensitive to outliers and scale. The dataset is small, and the analyst wants to keep all records. What is the most appropriate approach?

   • AApply RobustScaler to both features, which uses the median and IQR, making it less sensitive to the salary outliers.
   • BApply z-score standardization to both features.
   • CRemove the salary outliers before scaling.
   • DApply min-max scaling to both features, which will compress the salary outliers.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A data analyst is cleaning a dataset of employee records. The 'employee_id' column has duplicate values, but the 'name' and 'department' columns differ for those duplicates (e.g., ID 101 has 'John Smith' in 'Sales' and 'John Smith' in 'Marketing'). The analyst needs to create a unique employee list. What is the most appropriate action?

   • ARemove all duplicate rows based on 'employee_id', keeping the first occurrence.
   • BMerge the duplicate rows by concatenating the department values.
   • CReplace the duplicate 'employee_id' values with new unique IDs.
   • DInvestigate the duplicates to determine if they are the same person with outdated department info or different people with a data entry error.
   Reveal Solution Discussion Select an answer first
4. 8foundation · easy

   Which of the following is a common scaling method other than min-max normalization?

   • AOne-hot encoding
   • BPrincipal component analysis (PCA)
   • CImputation
   • DZ-score standardization
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   What is the purpose of filtering invalid entries in a dataset?

   • ATo sort the data in ascending order
   • BTo merge multiple datasets together
   • CTo remove rows that do not meet certain criteria or contain invalid values
   • DTo visualize the distribution of data
   Reveal Solution Discussion Select an answer first
6. 10expert · hard

   A data scientist is building a neural network. The dataset has a feature 'age' with a normal distribution (mean 40, std 10) and a feature 'income' with a right-skewed distribution (range $20k-$500k). The scientist wants to use min-max scaling on 'age' and z-score on 'income'. What is the most appropriate assessment of this approach?

   • AIt is acceptable, but using a consistent scaling method (e.g., both z-score) is often simpler and more interpretable.
   • BIt is incorrect because min-max scaling is never appropriate for neural networks.
   • CIt is incorrect because all features must be scaled using the same method.
   • DIt is correct because min-max is always better for skewed data and z-score is always better for normal data.
   Reveal Solution Discussion Select an answer first