我有一个使用DCC GARCH fit建模的双变量时间序列(你可以在这里找到data)。当我绘制Conditional Correlation Forecast
时,会抛出错误
> head(d_1)
x vibration_x Speed
1 2017-05-16 17:53:00 -0.132 421.4189
2 2017-05-16 17:54:00 -0.296 1296.8882
3 2017-05-16 17:56:00 -0.736 1254.2695
4 2017-05-16 18:00:00 -0.044 1209.6681
5 2017-05-16 18:01:00 -0.516 1212.5668
6 2017-05-16 18:02:00 0.492 1205.6841
garch11.spec.b = ugarchspec(mean.model = list(armaOrder = c(1,1)),
variance.model = list(garchOrder = c(1,1),
model = "sGARCH"), distribution.model = "norm")
dcc.garch11.spec.b = dccspec(uspec = multispec( replicate(2, garch11.spec.b) ), dccOrder = c(1,1), distribution = "mvnorm")
fit.b = dccfit(dcc.garch11.spec.b, data = d_1[,c(2,3)], fit.control = list(eval.se=T))
#Forecast
dcc.focast.a=dccforecast(fit.b, n.ahead = 100, n.roll = 0)
> plot(dcc.focast.a)
Make a plot selection (or 0 to exit):
1: Conditional Mean Forecast (vs realized returns)
2: Conditional Sigma Forecast (vs realized |returns|)
3: Conditional Covariance Forecast
4: Conditional Correlation Forecast
5: EW Portfolio Plot with forecast conditional density VaR limits
Selection: 4
Error in .plot.dccforecast.4(x, series, ...) : Not a matrix.
此外,Conditional Covariance Forecast
图仍然保持不变。这个可以吗?我认为它会捕捉拟合模型的波动性。
非常感谢任何帮助
谢谢, d
答案 0 :(得分:0)
在估算之前为数据尝试as.matrix()