多变量的时间序列协方差矩阵 - R.

时间:2017-11-20 06:29:16

标签: r covariance

我想为蒙特卡洛模拟创建一个时间序列数据的协方差矩阵,但是这个矩阵在多个资产之间。

即。我不仅想知道X(t),X(t + 1),...,X(t + n)之间的协方差,还想知道X(t),Y(t),Y(t +)之间的协方差。 1)等。

有一种简单的方法可以在R中生成一个协方差矩阵,它是数据帧中每个元素之间的协方差吗?

由于

1 个答案:

答案 0 :(得分:0)

这会产生最多nl滞后的协方差:

set.seed(123)
c1 <- data.frame(matrix(rnorm(90),ncol=3))
nl=2 # number of lags
c1wlags=embed(as.matrix(c1),nl+1)
nams=colnames(c1)
ndfs=paste(rep(nams,nl),rep(1:nl,each=ncol(c1)),sep="t-")
colnames(c1wlags)=c(nams,ndfs)
cov(c1wlags)


> cov(c1wlags)
               X1          X2           X3       X1t-1       X2t-1        X3t-1       X1t-2       X2t-2       X3t-2
X1     1.02206766 -0.13725252  0.153975857 -0.04152948  0.02664973 -0.123077418 -0.16800464  0.26851379 -0.25855822
X2    -0.13725252  0.73845275 -0.213758648  0.09490394 -0.11391726 -0.032365562  0.01863737  0.02661447 -0.11817727
X3     0.15397586 -0.21375865  0.797592802 -0.44057820  0.05532353 -0.007062553 -0.15315625  0.24462652 -0.19898668
X1t-1 -0.04152948  0.09490394 -0.440578196  0.96044907 -0.13594448  0.101988693  0.01357564  0.02734684 -0.11077465
X2t-1  0.02664973 -0.11391726  0.055323528 -0.13594448  0.74666015 -0.206596930  0.10511073 -0.11827071 -0.03883939
X3t-1 -0.12307742 -0.03236556 -0.007062553  0.10198869 -0.20659693  0.758188218 -0.38571731  0.05271998 -0.00125309
X1t-2 -0.16800464  0.01863737 -0.153156254  0.01357564  0.10511073 -0.385717313  0.92688519 -0.14256158  0.08240385
X2t-2  0.26851379  0.02661447  0.244626522  0.02734684 -0.11827071  0.052719977 -0.14256158  0.74893806 -0.20346301
X3t-2 -0.25855822 -0.11817727 -0.198986683 -0.11077465 -0.03883939 -0.001253090  0.08240385 -0.20346301  0.76041872