摘要
为了提高区间时间序列的模型预测精度,提出一种改进ARIMA模型的方法。将二元与三元区间序列分别转换为含有等量信息的实数序列,结合灰色模型中的数据累加处理方法和ARIMA模型实现实数序列建模,还原处理得到区间预测序列。数据分析表明,当区间序列波动较小时,不进行数据累加处理就能得到较高精度的区间预测序列,而当区间序列波动较大时,数据累加处理方法消除了原数据的随机性,更好地挖掘了建模序列的规律,因而得到更高精度的预测序列。
In order to improve the accuracy of the interval time series in forecasting model,an improved ARIMA model is proposed.The binary and ternary interval time series are changed into real sequences which contain the equivalent information,and then the cumulative processing method of the gray model is combined with ARIMA model to realize real sequence prediction.Finally the interval prediction can be obtained through the restoring procedure.The data analysis shows that the interval prediction sequence with high precision can be got without the data accumulated processing when the fluctuation of the interval sequence is small.But when the fluctuation of the interval sequence is large,the processing method of data accumulation eliminates the randomness of the original series and can be better to learn about the regular pattern of modeling sequence,so that the prediction sequence with higher precision can be got.
出处
《桂林电子科技大学学报》
2017年第1期79-86,共8页
Journal of Guilin University of Electronic Technology
基金
国家自然科学基金(71561008)
广西自然科学基金(2014GXNSFAA118010)
广西教育厅科研项目(KY2015YB113)