The code we used in k$-$means alt to compute new cluster means is given by for j in range $($k$)$:

Maljl $-$ np$.$mean $($X$[$cluster$\_$assignment $==$ jJ$,$ axis$-0)$

Above code can give an error in the case that some clusters have zero assigned points. Update the code $($one line$)$ to catch this exception. Handle the exception by printing a message "warning. A cluster with

zero points