如何隔离这种方法的地块？

Question

我正在使用rugarch 包并安装了一个模型。现在我想查看输出并使用绘图功能。我的问题是，第 5 个图包含一些子图，这些子图绘制在一个设备中，但我想将每个子图绘制在一个设备中。我怎样才能做到这一点？作为一个例子，我给你一个代码示例，它使用了包的 sp500ret 数据：

代码：

library(rugarch)
data(sp500ret)

somemodel<-ugarchspec(variance.model = list(model = "sGARCH", garchOrder = c(2, 2)), 
mean.model = list(armaOrder = c(1, 1), include.mean = TRUE), 
distribution.model = "ged")

somefit<-ugarchfit(spec=somemodel,data=sp500ret)

rollingesti = ugarchroll(somemodel, sp500ret, n.start=500,
 refit.every = 100, refit.window = 'moving', window.size = 500, 
  calculate.VaR = FALSE, keep.coef = TRUE)

plot(rollingesti,which=5)

plot(rollingesti,which=5) 将多个图绘制到一个设备中，我想隔离它们。

所以我想将它们作为单个地块并更大，现在它们太小了，因为它们都被放入一个输出中。

Answer 1

您的示例不起作用（至少对我而言），即它不会收敛。但是，这个可行：

library(rugarch)
data(sp500ret)
spec <- ugarchspec(distribution.model = "std")
mod <- ugarchroll(spec, data = sp500ret[1:2000,], n.ahead = 1, 
                 n.start = 1000,  refit.every = 100, refit.window = "moving", 
                 solver = "hybrid", fit.control = list(),
                 calculate.VaR = TRUE, VaR.alpha = c(0.01, 0.025, 0.05),
                 keep.coef = TRUE)

首先，我们找到一个在plot(mod, which = 5) 中使用的方法。可以通过

getMethod("plot", c(x = "uGARCHroll", y = "missing"))

您对以下几行感兴趣

.intergarchrollPlot(x, choices = choices, plotFUN = paste(".plot.garchroll", 
            1:5, sep = "."), which = which, VaR.alpha = VaR.alpha, 
            density.support = density.support, ...)

其中choices 是"Fit Coefficients (with s.e. bands)"。通过检查rugarch:::.intergarchrollPlot，我们终于到达rugarch:::.plot.garchroll.5。这些图不会在任何列表或类似列表中返回，因此我提供了一些修改后的版本，以便您可以单独使用它们。这里我改了前两行和最后一行：

library(xts)
x <- mod
vmodel = x@model$spec@model$modeldesc$vmodel
if (!x@model$keep.coef) 
  stop("\n\nplot-->error: keep.coef set to FALSE in estimation\n")
coefs = x@model$coef
m = dim(coefs[[1]]$coef)[1]
N = length(coefs)
Z = matrix(NA, ncol = m, nrow = N)
Zup = matrix(NA, ncol = m, nrow = N)
Zdn = matrix(NA, ncol = m, nrow = N)
for (i in 1:m) {
  Z[, i] = sapply(coefs, FUN = function(y) y$coef[i, 1])
  Zup[, i] = Z[, i] + sapply(coefs, FUN = function(y) y$coef[i, 
                                                             2])
  Zdn[, i] = Z[, i] - sapply(coefs, FUN = function(y) y$coef[i, 
                                                             2])
}
dt = sapply(coefs, FUN = function(y) as.character(y$index))
cnames = rownames(coefs[[1]]$coef)
np = rugarch:::.divisortable(m) # added rugarch:::

这是一个单独用于每个绘图的函数，i 是图形的编号，例如在这种情况下从 1 到 7：

plotFun <- function(i){
  plot(xts(Z[, i], as.POSIXct(dt)), type = "l", 
       ylim = c(min(Zdn[, i]), max(Zup[, i])), ylab = "value", xlab = "", main = "", 
       minor.ticks = FALSE, ann = FALSE, auto.grid = FALSE)
  lines(xts(Zdn[, i], as.POSIXct(dt)), col = 2)
  lines(xts(Zup[, i], as.POSIXct(dt)), col = 2)
  title(cnames[i], line = 0.4, cex = 0.9)
  grid()
}

例如：

plotFun(1)
plotFun(2)