scikit-learn: The Standard Workflow

fit and predict, train/test splits, scaling without leaking, the metrics that actually describe a classifier, pipelines, and cross-validation.

Learning Objectives

• →Fit a model and predict with it, getting the input shapes right
• →Split data into train and test, and say why the split has to exist
• →Fit a scaler on training data only, and explain what leaks if you do not
• →Read accuracy, precision and recall, and say when accuracy misleads
• →Chain preprocessing and a model into a Pipeline
• →Cross-validate, and read the spread of the fold scores as well as the mean

Coding Problems (6)