(a) Profiles of ridge coefficients for the prostate cancer example vs bound $B$ on $\ell _2$ norm of ${{w}}$, so small $B$ (large $\lambda$) is on the left. The vertical line is the value chosen by 5-fold CV using the 1 standard error rule. Adapted from Figure 3.8 of \citep{HastieBook}. Generated by ridgePathProstate.ipynb . (b) Same as (a) but using $\ell _1$ norm of ${{w}}$. The x-axis shows the critical values of $\lambda =1/B$, where the regularization path is discontinuous. Adapted from Figure 3.10 of \citep{HastieBook}. Generated by lassoPathProstate.ipynb .