This document discusses using various undersampling techniques to address class imbalance in a dataset containing two classes with uneven distributions. It generates an imbalanced dataset and fits a logistic regression model to visualize the decision boundary. It then applies several undersampling algorithms like RandomUndersampling, NearMiss versions 1-3 to balance the classes, fits new logistic regression models and compares the results. Precision, recall and F1 scores are calculated on the test set to evaluate model performance with different undersampling methods.