id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
dem-1058	Piłatowska, Mariola	Information and Prediction Criteria in Selecting the Forecasting Model	2011	20	.pdf	application/pdf	4933	171	49	Version II n=40 n=60 n=80 n=40 n=60 n=80 ARIMA vs. AR AIC AR AR AR AR AR AR BIC AR AR AR AR AR AR APE_SE ARIMA ARIMA AR ARIMA ARIMA AR APE_AE ARIMA AR ARIMA ARIMA ARIMA AR ARIMA vs. RW AIC ARIMA ARIMA ARIMA ARIMA ARIMA ARIMA BIC ARIMA ARIMA ARIMA RW RW ARIMA APE_SE ARIMA ARIMA ARIMA ARIMA ARIMA ARIMA APE_AE ARIMA ARIMA ARIMA ARIMA ARIMA ARIMA AR vs. RW AIC AR AR AR AR AR AR BIC AR AR AR RW RW AR APE_SE AR AR AR RW AR AR APE_AE AR AR AR AR AR AR Out-of-sample evaluation of IP forecasts (i.e. in the period 2011:01−2011:06) has been realized by the measures of accuracy (MSE, RMSE, U, MAPE) Accuracy measures for one-step-ahead forecasts of IP from different models in the period 2011:01−2011:06 − version II Accuracy measures n=40 n=60 n=80 ARIMA AR RW ARIMA AR RW ARIMA AR RW RMSE 2916.6 2908.4 3411.8 2716.42 2863.65 3386.73 2787.6 2697.1 3395.6 U 0.731 0.727 1.000 0.643 0.715 1.000 0.674 0.631 1.00 MAPE (%) 2.390 2.460 3.510 2.12% 2.210 3.500 2.220 1.990 3.520 For the window of 60 and 80 observations the ARIMA model and AR model give the smallest prediction errors.	cache/dem-1058.pdf	txt/dem-1058.txt