Chapter 2 · Machine learning basics
Models and parameters
A model is a formula with adjustable knobs (parameters); learning means finding good knob settings.
Preparing models and parameters
Chapter 2 · Machine learning basics
A model is a formula with adjustable knobs (parameters); learning means finding good knob settings.
The formula stays the same. Only the two knob settings change, and the line moves with them.
For a line, the formula is answer = slope × input + intercept. Feed in a number, get a prediction out.
The dots are the data. Turn the two knobs until the line runs through them, then press Fit for me to see the best settings.
answer = 0 × input + 8
Keep turning: the miss number shrinks as the line fits better. Real training finds the best settings with small automatic steps, which later lessons show.
What is a parameter?
What changes while a model learns?
A line has a slope and an intercept. How many parameters?