Train/Test Split Hold some data back to check yourself
A model that's only ever tested on data it already saw during training tells you nothing about how it'll perform in the real world. Splitting keeps a slice honest.
Fit a Model Learn a pattern from the training rows
Fitting a model means letting it find the relationship between your features and target on the training data — the same three lines, whichever algorithm you choose.
Evaluation Metrics Different metrics, different questions
Accuracy alone can lie to you, especially on imbalanced data. Each metric below answers a slightly different question about how the model is doing.
Confusion Matrix Where exactly the model gets it wrong
A confusion matrix breaks predictions into four buckets — it shows not just how often the model is wrong, but which kind of mistake it's making.
Feature Scaling Put every column on the same footing
Many models are sensitive to scale — a column measured in thousands can silently dominate one measured in single digits. Scaling levels the playing field first.