Instead of using VAR method, can I create separate model for each timeseries(TS) data.
In each model having on of the TS variable as target and other TS variables as features and using the lag of those features? and can I use MLP to achieve this VAR method?
1 post - 1 participant