Implement sqsplit

def sqsplit$($xTr$,$ yTr$)$:

$"""$

Finds the best feature, cut value, and impurity for a split of $($xTr$,$ yTr$)$ based on squared loss impurity.

Input:

xTr: n x d matrix of data points

yTr: n$-$dimensional vector of labels

Output:

feature: index of the best cut's feature $($keep in mind this is $0-$indexed$)$

cut: cut$-$value of the best cut

bestloss: squared loss impurity of the best cut

$"""$

n$,$ d $=$ xTr$.$shape

assert d $>0$ # must have at least one dimension

assert n $>1$ # must have at least two samples

bestloss $=$ np$.$inf

feature $=$ np$.$inf

cut $=$ np$.$inf

# YOUR CODE HERE

raise NotImplementedError$()$

#return feature, cut, bestloss