[EMAIL PROTECTED] (C. Cheung) wrote in message news:<[EMAIL PROTECTED]>... > Can MAPE be used as a criterion to choose the most suitable model for ARIMA/SARIMA?
If you are referring to a MAPE for the fitted values ....generally yes as it is one of a number of possible criteria ....such as variance, R-squared , AIC , BIC et. al. If you have a holdout sample then the problem becomes one where the "BEST MODEL" depends on how many values you withheld. Thus the tail wags the dog ! For more on model selection criteria and approaches to automatic model selection, you might read .... http://www.autobox.com/pdfs/catchword.pdf Dave Reilly Automatic Forecasting Systems http://www.autobox.com . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
