Building a large-scale binary classifier with hundreds or thousands of features for Google Display ads performance prediction.

Does logistic regression typically underfit or overfit in this setting? Describe the conditions that drive each, how you would detect the problem, and the techniques you would use to address it.

Cover regularization strength, feature selection, high-dimensional sparsity, learning curves, cross-validation.

Google Data Scientist interview question: Detect Overfitting or Underfitting in Logistic Regression Models. {"blocks": [{"key": "bb35fca0", "text": "Scenario", "type": "header-two", "depth": 0, "inlineStyleRanges": [], "entityRanges": [], "data": {...