Latent

Fit a line

Drag a line through ice cream sales data, score it by its squared misses, then watch gradient descent find the best line.

NoviceExplained in Getting less wrong

Fit a line to the ice cream sales

Each dot is one day: the temperature and how many cones sold. Drag the two handles to place a line that predicts sales from temperature.

Your line's typical miss65 cones
Best possible18 cones
Predicted sales at 30°C146 cones

Try this

  • Drag the handles to beat a typical miss of 25 cones, then 20.
  • Make the line slope downward. How big do the squares get?
  • Press Let the computer fit it from a few different starting lines. Does it always end in the same place?

Try "embedding", "softmax", "overfitting", or "backpropagation".