论文部分内容阅读
基于2008~2015年全国软饮料总产量的时间序列数据,采用ARIMA乘积季节模型对我国的软饮料总产量进行预报.首先使用差分的方法对序列作平稳化处理;其次,通过自相关与偏相关分析,结合模型选择的评估标准,建立ARIAMA乘积季节模型为ARIMA(4,2,3)×(1,1,1)12;然后运用最小二乘法估计模型的参数并对模型进行适应性检验,模型检验通过.利用建立的乘积季节模型ARIMA(4,2,3)×(1,1,1)12对2016年第一季度的软饮料总产量进行短期预测,平均相对误差珋e=0.0593,说明模型预测效果较好.结果表明,利用ARIMA乘积季节模型可以较好地对我国软饮料的总产量进行预报,具有一定的参考和应用价值.
Based on the time-series data of the national total soft drink production from 2008 to 2015, the ARIMA product seasonal model was used to forecast the total output of soft drinks in our country.Firstly, the differential method was used to stabilize the sequence.Secondly, through the autocorrelation and partial correlation analysis, Combined with the evaluation criteria of model selection, ARIAMA (4,2,3) × (1,1,1) 12 is established as the ARIAMA product seasonal model. Then the parameters of the model are estimated by using the least square method and the model is tested for fitness, model test Through the use of established product season model ARIMA (4,2,3) × (1,1,1) 12 for the first quarter of 2016, the total output of soft drinks short-term forecast, the average relative error 珋 e = 0.0593, indicating that the model prediction The result shows that using ARIMA product seasonal model can predict the total output of soft drinks in our country well and has some reference and application value.