从Auto arima模型解释日期

时间:2015-06-30 20:24:33

标签: r forecasting

以下是我的代码,

auto<-auto.arima(x)
auto_for<-forecast(auto,h=30)

> auto_for$x
Time Series:
Start = 1 
End = 74 
Frequency = 1 
 [1]  151  151  151  151  151  219  465  465  465  465  465  743  743  743  743  743  743  743  743  743  743  743
[23]  743  743  743  743  743  743  743  829  829  829  829  829  829 1004 1004 1004 1424 1424 1424 1822 1941 1941
[45] 1941 1941 1941 1941 1941 2076 2076 2252 2252 2252 2252 2252 2252 2252 2252 2252 2252 2252 2252 2940 2940 2940
[67] 2940 2940 3134 3134 3134 3207 3207 3465

> auto_for
    Point Forecast    Lo 80    Hi 80    Lo 95    Hi 95
 75       3510.397 3359.577 3661.217 3279.738 3741.056
 76       3555.795 3342.503 3769.086 3229.594 3881.996
 77       3601.192 3339.964 3862.419 3201.679 4000.705
 78       3646.589 3344.949 3948.229 3185.271 4107.907
 79       3691.986 3354.743 4029.230 3176.217 4207.755
 80       3737.384 3367.952 4106.815 3172.387 4302.380
 81       3782.781 3383.749 4181.812 3172.515 4393.047
 82       3828.178 3401.595 4254.761 3175.776 4480.580
 83       3873.575 3421.116 4326.035 3181.599 4565.552
 84       3918.973 3442.039 4395.907 3189.565 4648.380
 85       3964.370 3464.157 4464.582 3199.361 4729.379
 86       4009.767 3487.312 4532.222 3210.741 4808.793
 87       4055.164 3511.376 4598.953 3223.512 4886.817
 88       4100.562 3536.246 4664.878 3237.515 4963.608
 89       4145.959 3561.836 4730.081 3252.621 5039.297
 90       4191.356 3588.077 4794.635 3268.720 5113.992
 91       4236.753 3614.908 4858.599 3285.722 5187.785

我有预测值,但我无法从模型中获取日期。图表中没有日期,它已从0更改为91,而不是我的实际日期。我在开始时使用了xts变量。

更新

> a<-ts(ana)
> a
Time Series:
Start = 1 
End = 68 
Frequency = 1 
   final.day final.cumsum135
 1     16535             318
 2     16536             318
 3     16537             318
 4     16538             318
 5     16539             318
 6     16540             318
 7     16541             318
 8     16542             318
 9     16543             318
10     16544             318
11     16545             318
12     16546             318
13     16547             318
14     16548             318
15     16549             318
16     16550             318
17     16551             318
18     16552             318
19     16553             318
20     16554             318
21     16555             318
22     16556             318
23     16557             318
24     16558             318
25     16559             318
26     16560             369
27     16561             369
28     16562             369
29     16563             369
30     16564             369
31     16565             369
32     16566             369
33     16567             369
34     16568             369
35     16569             369

> auto<-arima(a)
Error in arima(a) : only implemented for univariate time series

我有什么办法可以在这里找回日期吗?

1 个答案:

答案 0 :(得分:0)

惠特日常系列,有时适合和预测&#34;失去&#34;日期。您可以使用索引手动获取日期:

y=x # x is your xts series
n=length(y)
model_a1 <- auto.arima(y)
# the plot
plot(x=1:n,y,xaxt="n",xlab="")
axis(1,at=seq(1,n,length.out=20),labels=index(y)[seq(1,n,length.out=20)],
     las=2,cex.axis=.5)
lines(fitted(model_a1), col = 2)
#the forecast
auto_for<-forecast(model_a1,h=30)
fcs=xts(auto_for$mean,seq.Date(as.Date(index(y)[n]),by=1,length.out=30))
fcs