Trace training versus prediction

Trace the path from training examples to a learned pattern, then from a new example to a prediction. You’ll keep training and prediction separate so it is clear when the model is learning and when it is being used.