Implement sqimpurity

First, implement the function sqimpurity, which takes as input a vector $$

of $$

labels and outputs the corresponding squared loss impurity:

$=1()2,$ where $==1.$

Again, the squared loss impurity works fine even though our final objective is classification. This is because the labels are binary and classification problems can be framed as regression problems.

def sqimpurity$($yTr$)$:

$"""$

Computes the squared loss impurity $($variance$)$ of the labels.

Input:

yTr: n$-$dimensional vector of labels

Output:

squared loss impurity: weighted variance$/$squared loss impurity of the labels

$"""$

N$,=$ yTr$.$shape

assert N $>0$ # must have at least one sample

impurity $=0$

# YOUR CODE HERE

raise NotImplementedError$()$

return impurity