#diagnostics

Question 1: Select all that apply

mediumAI-checkedLinear Models & Regularization

Each plot shows the training prediction errors $y - \hat y$ against the input $x$ of a model fitted by least squares. The linear model is $\hat y = w_0 + w_1 x$ and the quadratic model is $\hat y = w_0 + w_1 x + w_2 x^2$ . Which statements are true? Select all that apply.

Three scatter plots of prediction error against x on [-1, 1]. A: points spread evenly around 0. B: a U-shaped band that dips to about -0.35 at x = 0 and rises to about 0.65 at x = ±1. C: a U-shaped band from about 0 at x = 0 up to about 1 at x = ±1.

B could be the errors of the quadratic model1
A could be the errors of the quadratic model2
C could be the errors of the quadratic model3
B could be the errors of the linear model4
C could be the errors of the linear model5

Timed test