Open RobertoDF opened 2 months ago
Yes, this happens occasionally. The reason it happens is if during a random split of the data one of the cross validation chunks happens to not have any timepoints that contain a specific model regressor (like no omissions, or no changes, or no licks), then the model is singular and we can't fit the model. In my hands, this happens pretty infrequently, so I simply ignored the sessions where it happened. It also tends to happen the most in sessions where the mice lick very little, so I would probably exclude those sessions anyways. If you set the cross validation to be larger (like 10x validation), it should happen even less frequently.
Hi, I run in the following error during cross validation: