模式匹配在GDP时间序列预测中的应用

深入分析了自回归整合平均移动模型(ARIMA)以及时间序列在动态时间弯曲(DTW)距离下的匹配技术，并在此基础上建立了由多条时间序列集合而成的GDP整合时间序列预测模型。其基本思想是对与目标序列有着相同发展趋势的时间序列进行搜寻、匹配，并对这些具有相同历史进程时间序列所蕴含的信息进行充分的挖掘与利用，整合形成新的时间序列预测模型。仿真实验表明：整合GDP预测模型的预测准确率显著高于普通ARIMA模型的预测准确率，从而证实了整合时间序列模型用于GDP预测的准确性。

GDP Time Series Forecasting Based on Pattern Matching

Based on the analysis of autoregressive integrated moving average model(ARIMA) and time series pattern matching method under dynamic time warping distance(DTW), this paper presents an integrated method combining several other time series to certain GDP time series forecasting. For time series with same development trend in distinct backgrounds, we firstly applying pattern matching method to search for the optimal matched time periods according to object time series and then combining these time series together for a better forecasting. The consequences of GDP forecasting show that this integrated model results in greater forecasting accuracy comparing to single ARIMA model.