|
|
楼主
楼主 |
发表于 2011-5-10 20:13:47
|
只看该作者
月度数据预测(急)
data sample;
input x dollar12.;
time=intnx('month','01JAN2009'd,_n_-1);
datalines;
$294,611.00
$618,988.10
$1,107,091.40
$482,419.60
$792,156.80
$703,888.90
$427,747.00
$582,007.30
$530,385.48
$511,895.00
$1,118,648.50
$868,240.69
$837,273.00
$802,946.00
$627,884.10
$558,911.80
$1,580,109.08
$962,075.29
$797,031.86
$974,035.40
$900,323.24
$650,623.00
$723,836.00
$2,195,369.95
$1,646,890.08
$1,626,268.44
$1,980,308.69
$1,677,193.00
;
proc gplot data=sample;
plot x*time;
symbol c=black ;
run;
proc arima data=sample;
identify var=x nlag=12 minic p=(0:5) q=(0:5);
estimate p=1 q=1;
forecast lead=5 id=time out=result;
run;[/code]
我用上面的代码得到的结果，相当不理想，区间相当大，没有啥现实意义。简单标准化之后，得到的结果也不理想。
在这里请教大牛，如何改进。思路，代码都可以，小弟在先谢过！！ |
|
