NCA-GENL Question 35
Single answerYou are building a machine learning model to predict customer churn for a subscription-based service. After preparing the dataset, you decide to validate the model's performance using cross-validation. Which of the following is the primary benefit of using cross-validation in this scenario?
- A
It helps prevent the model from underfitting the training data.
- B
It ensures the model is evaluated on multiple subsets of data, reducing the risk of overfitting to a specific training set.
- C
It increases the size of the training dataset by combining test and validation data.
- D
It automatically selects the best set of features for the model.
Show answer and explanation
Correct answer: B
Explanation
Cross-validation is a robust technique for model evaluation. It splits the dataset into multiple subsets (folds) and trains the model on some while validating it on others, rotating through all folds. This process ensures the model generalizes well by providing performance metrics averaged over multiple training-validation splits, reducing the bias or variance associated with evaluation on a single partition.
- A. Incorrect.
Preventing underfitting is not a primary purpose of cross-validation. Underfitting is typically addressed by improving model complexity or feature engineering.
- B. Correct.
This is correct. Cross-validation divides the dataset into multiple subsets (folds), ensuring the model is evaluated on different partitions of the data. This reduces the risk of overfitting to a single training set and provides a more reliable estimate of model performance.
- C. Incorrect.
Cross-validation does not increase the size of the training dataset. It uses the same amount of data but splits it into different subsets for training and validation.
- D. Incorrect.
Cross-validation does not perform feature selection. Feature selection is a separate process that involves identifying the most relevant variables for the model.