Hands-On Exercise: Data Preparation for ML CT-AI Practice Questions (Page 2)

Questions 6–10

1. 6application · medium

   A team is building a model to predict employee attrition. The 'department' column has 10 unique values, and the 'salary_level' column has values 'low', 'medium', and 'high'. They plan to use a tree-based model (random forest). Which encoding strategy is most appropriate?

   • ALabel encode both 'department' and 'salary_level'.
   • BOne-hot encode both 'department' and 'salary_level'.
   • COne-hot encode 'department' and ordinal encode 'salary_level'.
   • DOrdinal encode 'department' and one-hot encode 'salary_level'.
   Reveal Solution Discussion Select an answer first
2. 7expert · hard

   A data scientist is preparing a dataset for a binary classification task. The 'age' column has outliers (e.g., 150, 200) and the 'income' column has 5% missing values. They plan to use a tree-based model (random forest). Which approach is most appropriate?

   • AApply StandardScaler to both features and impute income with 0.
   • BRemove rows with age outliers and impute missing income with the mean.
   • CLeave age outliers as-is and drop rows with missing income.
   • DCap age outliers at a reasonable maximum (e.g., 100) and impute missing income with the median.
   Reveal Solution Discussion Select an answer first
4. 8expert · hard

   An ML engineer is preparing a dataset for a k-nearest neighbors (kNN) model. The dataset contains a 'salary' feature with a right-skewed distribution and a 'years_experience' feature that is roughly normal. The salary feature has a few extreme high values. Which preprocessing approach is most appropriate for kNN?

   • AApply log transformation to 'salary' and then StandardScaler to both features.
   • BApply MinMaxScaler to both features without any transformation.
   • CRemove the extreme salary values and apply RobustScaler to both features.
   • DApply StandardScaler to both features without any transformation.
   Reveal Solution Discussion Select an answer first
5. 9foundation · easy

   Which technique is commonly used to handle missing values in a dataset?

   • AImputation with the mean or median
   • BPrincipal component analysis (PCA)
   • COne-hot encoding
   • DData augmentation
   Reveal Solution Discussion Select an answer first
6. 10expert · hard

   A data science team is building a model to predict loan default. They have a dataset with 1 million rows and 200 features. The team has limited compute resources and needs to train the model within a strict time budget. They also need to ensure the model's performance estimate is reliable. Which approach best balances these constraints?

   • AUse all 200 features and train a deep neural network with a 70/30 train/test split.
   • BSplit into train/test, then apply PCA to reduce dimensionality, and train a random forest.
   • CRandomly sample 10% of the data, split into train/test, and train a gradient boosting model.
   • DPerform feature selection using L1 regularization, then split into train/validation/test and train a logistic regression model.
   Reveal Solution Discussion Select an answer first