改进的ARIMA模型预测精度分析

Analysis of prediction accuracy of improved ARIMA model

经典的时间序列ARIMA模型因为精度准确经常被用于预测,以此模型为基础尝试提高该模型的预测精度,ARIMA模型在预测中都只是对某个时间点进行的研究,然而有时对未来的影响不仅是一个点的效应,更是一段时间积累而导致最后结果变化.所以我们不访考虑一段时间上对ARIMA模型进行改良,在原模型基础上建立改进模型,并分析改进后模型的预测精度,将模型改进前后以及应用指数平滑法的预测结果进行拟合度和精度的比较,从模型拟合程度、精度度量指标对比分析改进前后时间序列ARIMA模型和指数平滑预测模型,结果显示改进后时间序列ARIMA模型预测最为精确,其次为指数平滑法的预测模型,最后为改进前时间序列ARIMA模型预测.证明了改进后模型的精度确有所提高.

The classic ARIMA model of time series is often used for forecasting because of its accuracy. Based on this model,this paper tried to improve the forecasting accuracy of this model. ARIMA model only studies a certain time point in forecasting,but sometimes its influence on the future is not only the effect of a point,but also the change of the final result caused by a period of accumulation. This paper considered improving ARIMA model for a period of time,build an improved model based on the original model,analyze the prediction accuracy of the improved model,compare the fitting degree and accuracy of the prediction results before and after the model improvement and using exponential smoothing method,and compare and analyze the time series ARIMA model and exponential smoothing prediction model before and after the improvement from the model fitting degree and accuracy metrics.The results showed that the improved time series ARIMA model was the most accurate,followed by exponential smoothing prediction model,and finally the time series ARIMA model before the improvement. It was proved that the accuracy of the improved model was improved.