Hello everyone,
I’ve been stuck with this problem.
If your train_val_split function implementation is correct, please click my name and message your notebook as an attachment.
Your notebook is valid. Odds are good that you submitted a notebook without saving all the changes.
Please submit the notebook you sent via direct message. The score should be 100.
Thanks @balaji.ambresh 
Hi, I’m facing a similar issue, and I tried restarting the kernel + saving the checkpoint and then submitting it for grading, and still am being awarded 0 points for the train_val_split function (everything else is correct acc. to the grader)
Please click my name and message your notebook as an attachment.
Sorry that the grader feedback wasn’t helpful. I’ve notified the staff to fix this issue and explain the test on this thread.
This is the test case that fails in your case:
time = [1, 2, 3, 4, 5]
time.shape=(5,)
series.shape=(5,)
time_step=2
np.where(time==time_step)= (array([1]),)
As you can see, when time doesn’t start at 0 there’s a problem:
>>> time = [1, 2, 3, 4, 5]
>>> time[:1] # Not what we want
[1]
We want [1, 2]
Thanks for the clarification. I thought we wanted the slice to happen at time = 1100 days, not at the 1100th timestep, that’s why I used the np.where method.