摘要
对比传统时间序列分析模型(线性回归、二次滑动平均、一次平滑、二次指数平滑和三次指数平滑等)与ARIMA模型在山东省粮食总产量中的拟合精度,并应用ARIMA(2,1,12)模型预测了未来3年内山东省粮食总产量。结果表明,在山东省粮食总产量拟合中,ARIMA(2,1,12)模型得到的粮食总产量拟合值与观测值的相对误差处于±10%和±5%范围内的分别为73.333%和53.333%,回归方程的决定系数为0.959,优于传统时间序列分析模型;利用ARIMA(2,1,12)模型预测未来3年内山东省粮食总产量,粮食总产量有逐年上升的趋势,且增长率逐年上升。
The classical time series analysis model （linear regression, two step moving average, one step smoothing, two step EXSMOOTH, three step EXSMOOTH, etc.） and the ARIMA model were compared to predict total corn yield, and ARIMA （2,1,12） model was applied to predict the total corn yield in future 3 years. Results showed that the ARIMA （2,1,12） model was better than the classical time series analysis model in total corn yield of Shandong Province. The determine coefficient of regressive equation was 0. 959 and the relative error between fitted value and measured value among 73. 333% and 53. 333% were ±10% and ±5%, respectively. The total corn yield and the increasing ratio ascended year by year in future 3 years with predicting model of ARIMA（2,1,12） applied.
出处
《水土保持研究》
CSCD
北大核心
2007年第3期309-311,共3页
Research of Soil and Water Conservation
基金
鲁东大学大学生科技创新基金资助
