On Sat, 21 May 2011, B�lent Miran wrote: > During my ARIMA modellings, I encountered several bugs. For > example when I got a trial with certain p and q parameters, I > couldn't find the ARIMA estimation with those parameters. > Sometimes I used the clear button to clean the previous > model's parameters, but the new ARIMA estimation was presented > with previous model's parameters. It may be a bug, or not. Perhaps the algorithm just failed to converge. Could you send your data so that I can replicate your problem?