Start by defining classification and regression in terms of their output types and learning objectives. Then contrast them using examples, evaluation metrics, and algorithms, and finally discuss how the choice depends on the problem and business context, especially at Netflix.
Pro tip: Emphasize that the distinction is not just about the output type but also about the loss functions and evaluation metrics, and mention that some problems can be framed either way depending on the business need.
Explain that classification predicts discrete class labels while regression predicts continuous numerical values.
Provide concrete examples for each, such as predicting whether a user will click (classification) vs. predicting the number of hours they will watch (regression).
Mention that some algorithms are inherently for classification (logistic regression, decision trees for classification) and others for regression (linear regression, regression trees), while some can do both (random forests, neural networks).
Contrast metrics: accuracy, precision, recall, F1, AUC-ROC for classification; MSE, MAE, RMSE, R-squared for regression.
Discuss how the choice impacts decision-making and how sometimes a regression output can be thresholded to make a classification, or vice versa, depending on the problem.
AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.