I’m not saying anything about metrics. The format of the input files is different from what that notebook expects. There was a time in the old format that the input files had 51 or 52 columns, with t_id, 50 features, and a target (in the training file). That is no longer the case. So if you’re just trying to run that notebook as is, it won’t be removing the id column (because the column header changed) and it will be incorrectly leaving in several additional columns (data_type, era) so I wouldn’t expect anything downstream of loading the data to work at all.
If you have changed those things and are still asking for advice,then you need to provide a link to the file you’re actually using, not some outdated file that doesn’t match what you’re using.