时间序列论文.docx

时间序列分析大论文课题：基于ARIMA模型的金城股份 (000820)收盘价的时间序列分析及预测。胡毅翔,学号:1322010036基于ARIMA模型的金城股份 (000820)收盘价的时间序列分析及预测。一.摘要本文用时间序列分析方法，对一段时间序列进行了拟合。通过对2013/08/29至2013/11/29金城股份 (000820) 收盘价序列进行观察分析，建立合适的ARIMA模型，对未来1天的金城股份 (000820) 收盘价序列进行预测。然后对预测值和真实值进行比较，得出结论，所建立的模型有较好的拟合效果，从而提供了一个行情预测的有效方法。关键词：时间序列 收盘价ARIMA 时间序列分析 预测 二.数据预处理及具体模型（建模）1.数据录入：2013/08/29至2013/11/29金城股份 (000820) 收盘价时间收盘时间收盘时间收盘2013/08/2910.212013/09/3010.442013/11/0410.462013/08/309.852013/10/0810.32013/11/0510.32013/09/0210.262013/10/0910.362013/11/0610.512013/09/0310.222013/10/1010.372013/11/0710.12013/09/0410.072013/10/1110.492013/11/089.772013/09/059.832013/10/1410.292013/11/119.762013/09/069.942013/10/1510.552013/11/129.782013/09/0910.172013/10/1610.692013/11/139.482013/09/1010.232013/10/1711.342013/11/149.522013/09/119.862013/10/1811.112013/11/159.72013/09/129.72013/10/2111.432013/11/189.772013/09/139.992013/10/2211.712013/11/1910.072013/09/169.82013/10/2311.62013/11/2010.042013/09/1710.292013/10/2411.222013/11/2110.042013/09/1810.772013/10/2511.162013/11/2210.642013/09/2310.72013/10/2811.652013/11/2510.412013/09/2410.262013/10/2910.492013/11/2610.092013/09/2510.262013/10/3010.472013/11/2710.052013/09/269.92013/10/3110.422013/11/2810.162013/09/2710.062013/11/0110.42013/11/299.92绘制时序图程序如下：data a;input a@@;time=_n_;cards;10.219.8510.2610.2210.079.839.9410.1710.239.869.79.999.810.2910.7710.710.2610.269.910.0610.4410.310.3610.3710.4910.2910.5510.6911.3411.1111.4311.7111.611.2211.1611.6510.4910.4710.4210.410.4610.310.5110.19.779.769.789.489.529.79.7710.0710.0410.0410.6410.4110.0910.0510.169.9;Procgplot;plot a*time;symbol1v=star i=join c=black;run;原时序图3.一阶差分时序图通过原始数据的时序图可以明显看出，此序列非平稳，因而对序列进行一阶差分。data yx_51;input x@@;difx=dif(x);t=1+_n_-1;cards;10.219.8510.2610.2210.079.839.9410.1710.239.869.79.999.810.2910.7710.710.2610.269.910.0610.4410.310.3610.3710.4910.2910.5510.6911.3411.1111.4311.7111.611.2211.1611.6510.4910.4710.4210.410.4610.310.5110.19.779.769.789.489.529.79.7710.0710.0410.0410.6410.4110.0910.0510.169.9;pro