Hi everyone, when I try to pass exogenous variables to AutoARIMA, it gives me the following error: I tried dropping all the covariates but one for all the existing covariates, but the error persists. Any idea on how to solve this?

Are you including a trend column? That model builds one internally so you'd be introducing colinearity

or you have all zeros in one column? https://github.com/robjhyndman/forecast/issues/699

I don't have an all-zeros column, not sure about the trend one. I currently have these ones: ['ticker', 'datetime', 'open_lag12', 'high_lag12', 'volume_lag12', 'close_diff1_lag12', 'close_diff12_lag12']. But as I said, even if I keep only one of them, it still gives me the error

Seems like you're using a vscode notebook. When you hit that error you can type

%debug

in a separate cell and that will take you to the place where the error happens, you can then inspect the matrix

Ok, I restarted the kernel and ran everything again, had two highly correlated columns (97%), dropped one of them and it is working now.