01 / FIRST EXPLORATION
How can a line separate
two classes of points?
Move the line to separate two classes. The points are training data; the line is a simple classification model.
CLASS A CLASS B
EXPERIMENT 01 — LINEAR BOUNDARYYOUR MODEL
Move the line.
Separate the points.
Points above the line are predicted as A, and points below as B. Change the slope and height to change the predictions.
Correct on training data—
A white outline marks a point the current line misclassifies.
01
What just happened?
We use each point’s coordinates to decide which side of the line it falls on. That rule is a model; the slope and height are its parameters.
You moved the sliders yourself. What would a computer need to find good values on its own?
Next: What makes a boundary good? ↗THINK ABOUT IT
If the line classifies every training point correctly, will it always classify new points correctly? Training accuracy does not guarantee performance on unseen data.