Seasonal Time Series Data Forecasting by Using Neural Networks Multiscale Autoregressive Model | Science Publications

American Journal of Applied Sciences

A. J. Endharta¹  Suhartono¹  B. S.S. Ulama¹ 

关键词: Neural networks;    multiscale;    MODWT;    seasonal;    time series;    NN-MAR;   

DOI  :  10.3844/ajassp.2010.1372.1378

学科分类：自然科学（综合）

来源: Science Publications

【 摘 要 】

Problem statement: The aim of this research was to study further some latest progress ofwavelet transform for time series forecasting, particularly about Neural Networks MultiscaleAutoregressive (NN-MAR). Approach: There were three main issues that be considered further in thisresearch. The first was some properties of scale and wavelet coefficients from Maximal OverlapDiscrete Wavelet Transform (MODWT) decomposition, particularly at seasonal time series data. Thesecond focused on the development of model building procedures of NN-MAR based on the propertiesof scale and wavelet coefficients. Then, the third was empirical study about the implementation of theproposed procedure and comparison study about the forecast accuracy of NN-MAR to otherforecasting models. Results: The results showed that MODWT at seasonal time series data also hasseasonal pattern for scale coefficient, whereas the wavelet coefficients are stationer. The result ofmodel building procedure development yielded a new proposed procedure of NN-MAR model forseasonal time series forecasting. In general, this procedure accommodated input lags of scale andwavelet coefficients and other additional seasonal lags. In addition, the result showed that the proposedprocedure works well for determining the best NN-MAR model for seasonal time series forecasting.Conclusion: The comparison study of forecast accuracy showed that the NN-MAR model yields betterforecast than MAR and ARIMA models.

【 授权许可】

Unknown