I think that a good solution would be to add an extra conditional for when window_end == k-1, and assign the current sum to max_sum. With that you have a start point to compare and it should work even with negative numbers. You would need to modify the exisiting if stat...

No, the issue is that if your subarray is [-1] with window size 1, you're not going to get the correct solution because your average will be -1.

In that case, 0 always > -1. In python, should set window_sum to -sys.maxsize - 1

problem is not negative numbers, but the fact that we are intializing the max sum as 0, instead of doing that u can intialize that with the first window sum, that will work even in case of negative numbers

carlos solution is also good but it introduce one more condition

It could also be solved by initializing your sum/avg with least possible value of the type you chose.

For those who're struggling with this:

I code in C++, I was trying to initialize the double max_average by DBL_MIN since max_average is double. I turned out that I have to initialize it by INT_MIN instead.

The solution would look something like this:

class Solution ...