Hi, I am trying to use autoarima function to make prediction of more than 10000 items ( each has 100 values so it is 10000x100 dimension matrix) with frequency 52(weekly forecast). The problem is that some of fitted arimas have drift and some not. It gives this error and stop processing Best model: ARIMA(1,0,0)(0,1,0)[52] with drift Error in predict.Arima(fitted.arima, n.ahead = 28) : 'xreg' and 'newxreg' have different numbers of columns: 1 != 0 And when I put the newxreg=number of rows+number of predicts which is in my model 128 in the predict function it gives the same error for non drifts. Best model: ARIMA(2,0,0)(0,1,0)[52] Error in predict.Arima(fitted.arima, n.ahead = 28, 128) : 'xreg' and 'newxreg' have different numbers of columns: 0 != 1 Do any one has any recommendation to get rid of this error and could be able to compile my program and get prediction for my 10000 items? I will highly appreciate in advance. Regards Jeela [[alternative HTML version deleted]]