Validation and test sets are larger than the shorter time-series

[michelidiego]

Hi Everyone,

I'm trying to run a model where there are 4 different unique ids (one timeseries for each) and I'm 100% confident (I've triple checked) that each ID has at least 300 values. I'm using the long horizon model documentation and when I set the validation and test size , I use 20% so val_size and test_size are both 60. So far so good.

The issue appears when I try to use the cross_validation method and I get the following warning:

"Validation and test sets are larger than the shorter time-series."

After that, the function crashes and the following error shows:

negative dimensions are not allowed

My main question is, How can the validation and test sets be larger than the shorter time-series when I know for a fact the each of the 4 time series has 300 elements.

Thank you for your help