Answer: Penalize the classification, Resample the dataset using undersampling or oversampling, Generate synthetic samples in the minority class

The question asks for three methods to resolve class imbalance in a binary classification model. Option A (Penalize the classification) refers to cost-sensitive learning, where misclassification of the minority class is assigned a higher penalty, directly addressing imbalance. Option B (Resample the dataset using undersampling or oversampling) involves adjusting the class distribution by either reducing majority class samples (undersampling) or increasing minority class samples (oversampling). Option D (Generate synthetic samples in the minority class) uses techniques like SMOTE to create synthetic data for the minority class, balancing the dataset. These are established methods for handling imbalance. Option C (Normalize the training feature set) is unrelated to class imbalance, as it standardizes feature scales but does not address distribution skew. Option E (Use accuracy as the evaluation metric) is unsuitable because accuracy can be misleading with imbalanced data, as it may favor the majority class; metrics like F1-score or AUC-ROC are preferred. The community discussion, with 100% consensus on ABD and upvoted comments explaining why C and E are incorrect, supports this selection.

Author: LeetQuiz Editorial Team