How do class weights work in logistic regression, and when do you use them?

medium

Answer

$Class weights re-weight the log-loss contributions of each class so the minority class carries more penalty when misclassified.$
$\mathrm{weight}_{c}\; = \;n_{\mathrm{samples}}\; / \;(n_{\mathrm{classes}}\; \cdot \;n_{c})$
$Effect: pushes the decision boundary toward the majority class and improves recall on the minority class, at the cost of precision.$
$Use when the positive class is rare and false negatives are more costly than false positives.$

Check yourself — multiple choice

Class weights re-scale the features
They re-weight per-class contributions to the loss, boosting minority-class recall
They replace threshold tuning
They only apply to trees

$\mathrm{class}_{\mathrm{weight}}$ rescales the loss per class ⇒ better minority-class recall (usually) at some precision cost.

#logistic-regression#imbalance

215 interview questions in this topic.