Factors Involved in ML Algorithm Selection CT-AI Practice Questions (Page 4)

Questions 16–20

1. 16expert · hard

   A telecom company wants to predict customer churn. They have a dataset with 1 million records, but 30% of the feature values are missing. The data includes both numerical and categorical features. The model must be trained on a single machine with limited memory. Which algorithm is most suitable?

   • ALogistic regression with mean imputation
   • BK-nearest neighbors with imputation
   • CDeep neural network with imputation
   • DDecision tree with built-in handling of missing values
   Reveal Solution Discussion Select an answer first
2. 17application · medium

   A marketing analytics team is building a model to predict customer churn. They have a dataset with 10,000 records and 100 features. They noticed that a complex model performs very well on training data but poorly on validation data. The business needs a model that generalizes well to new customers. Which approach should they take?

   • AAdd more features to the model
   • BIncrease the complexity of the model
   • CUse a simpler model like logistic regression with regularization
   • DUse a deep neural network with more layers
   Reveal Solution Discussion Select an answer first
4. 18foundation · easy

   A dataset contains many missing values and outliers. How does this data quality issue affect algorithm suitability?

   • AIt may make algorithms sensitive to outliers, such as linear regression, less suitable without preprocessing
   • BIt has no effect because all algorithms handle missing data identically
   • CIt makes tree-based models always unsuitable
   • DIt only affects the visualization, not the algorithm choice
   Reveal Solution Discussion Select an answer first
5. 19foundation · easy

   Which data characteristic is most important when deciding between a simple model and a deep neural network?

   • AThe alphabetical order of column names
   • BThe file format of the data (CSV vs. JSON)
   • CThe volume of available training data
   • DThe color depth of images if any
   Reveal Solution Discussion Select an answer first
6. 20foundation · easy

   A regulatory requirement demands that loan decisions be explainable to customers. Which algorithm characteristic is most important in this scenario?

   • AMaximum predictive accuracy
   • BHigh scalability
   • CLow training time
   • DHigh interpretability
   Reveal Solution Discussion Select an answer first