Decision Trees DATA-SCIENCE-FOUNDATIONS Practice Questions (Page 5)

Questions 21–23

1. 21application · medium

   A data analyst is building a decision tree to predict equipment failure. The dataset includes a categorical feature 'maintenance_type' with values 'none', 'scheduled', and 'emergency'. The analyst wants to use this feature in the tree. Which approach is most appropriate for handling this categorical feature in a CART-style decision tree?

   • AUse the feature directly, allowing the tree to consider binary splits on category subsets.
   • BOne-hot encode the feature into binary columns.
   • CAssign numerical codes 1, 2, 3 to the categories.
   • DDrop the feature because decision trees cannot use categorical features.
   Reveal Solution Discussion Select an answer first
2. 22application · easy

   A data scientist has trained a decision tree to predict whether a customer will buy a product. The tree's root node splits on 'income > 50k'. The left branch (income ≤ 50k) leads to a leaf predicting 'no purchase', and the right branch (income > 50k) leads to a node that splits on 'age > 30'. What does the path 'income > 50k' → 'age > 30' → leaf 'purchase' represent?

   • AA decision rule that predicts purchase for customers with income above 50k and age above 30.
   • BThe feature importance of income and age.
   • CThe Gini impurity of the leaf node.
   • DThe training accuracy of the tree.
   Reveal Solution Discussion Select an answer first
3. 23foundation · easy

   Which approach is commonly used by decision tree algorithms to handle missing values during training?

   • AAssigning missing values to the most frequent category
   • BUsing surrogate splits that approximate the best split when a value is missing
   • CImputing the mean or median for all missing values
   • DDropping all rows with missing values
   Reveal Solution Discussion Select an answer first