$2\mathrm{.}1$ LASSO regression $[4$ points$]$

Recall that the loss function to be optimized under LASSO regression is:

$E_L={\displaystyle \underset{i=1}{\overset{n}{}}}(Y_i-(hat(w{)}_0+x_i(hat(w))){)}^2+||hat(w)|{|}_1$

where

$||hat(w)|{|}_1={\displaystyle \underset{i=1}{\overset{d}{}}}|hat(w{)}_i|$

and $$ is our regularization constant.

Suppose our $$ is much too small; that is$,$

${\displaystyle \underset{i=1}{\overset{n}{}}}(Y_i-x(hat(w)){)}^2+||w|{|}_1$~~${\displaystyle \underset{i=1}{\overset{n}{}}}(Y_i-x(hat(w)){)}^2$

How will this affect the magnitude of:

$($a$)$

The error on the training set?

$($b$)$

The error on the testing set?

$($c$)$ point there is problem upluding question