I got this error code when I wrote a first solution to the split_dataset(). I am not entirely sure what causes it.
I used this to try to iterate through all rows, I thought that was how it had been done earlier in the course.
Now I’ve fixed my code and everything works. But it like to know what’s happening here! 
This must be a simple error. I think you know your error, since you seemed to fix it.
" X.shape[0]", the number of samples, is equal or larger than “node_indices” which is the list of “active indices”. But, your old code always check all samples. In the error case, there are 11 samples in X. But, node_indices has only 7. As the result, you put 11 samples to either left or right. But, expected result is to put 7 samples to either left or right. Since the array size is different, a broadcasting error should be shown up.
Hope this clarifies.

I was about to write a reply, but @anon57530071 has given us a very good direction, so I would just to post the description of the exercise here which tells us which index we need to iterate through.
Good luck @Rickard
Raymond
thank you! Yeah I realised that I hadn’t read the instructions thoroughly enough… Now I get it!
1 Like