Title | ||
---|---|---|
Construction of an artificial neural network for simple exponential smoothing in forecasting |
Abstract | ||
---|---|---|
Time series models have been applied to forecast the market trends. Simple exponential smoothing (SES) method was one of them widely used as a forecasting tool in time series data. In this method, a smoothing parameter needs to be chosen in such a way tO minimize sums of squares forecast errors. By deriving the equivalence of the smoothing equation and the artificial neural network algorithm, we have shown that SES is equivalent to a special case of artificial neural network (ANN). Furthermore. we propose an adaptive simple exponential forecasting (ASES) method which merges SES and ANN to selfmodify the weighted connection and to obtain the estimate of the smoothing parameter for better forecasting. Both SES and ASES have been applied to the Standard and Poor 500 composite indexes of past twenty years in stock-market forecasting. ASES is generally superior to SES in forecasting. |
Year | DOI | Venue |
---|---|---|
1994 | 10.1145/326619.326755 | SAC |
Keywords | Field | DocType |
simple exponential smoothing,adaptive simple exponential smoothing,simple exponential,artificial neural network,time series model,exponential smoothing,time series data,sum of squares | Exponential smoothing,Applied mathematics,Mathematical optimization,Computer science,Smoothing,Artificial neural network,Additive smoothing | Conference |
ISBN | Citations | PageRank |
0-89791-647-6 | 1 | 0.48 |
References | Authors | |
3 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Doug W. Mahoney | 1 | 1 | 0.48 |
Ruey-Pyng Lu | 2 | 10 | 1.53 |
Shaun-Inn Wu | 3 | 18 | 2.95 |