CHAPTER 04 · THE REVEAL

The model kept the wrong features.

Regularisation fixed the prediction problem. Now for the invoice. Lasso promised to tell you which features mattered, so you pull up its top 20 — coloured by the role the codebook assigned in secret. Some of the survivors are spurious. You run a Random Forest on the same data and it picks a different 20. Two models, two stories, one dataset. Which do you publish?

loading importances …