具有特定 VAR 模型的预测滞后 [英] Forecasts with specific VAR model lags

问题描述

提前致谢.在这篇文章中，我问了如何在 VAR 模型中选择特定滞后的问题.在限制"和系数"函数的快速回复和信息之后，我能够成功运行具有我想要的特定滞后的 VAR 模型.但是，使用受限 VAR 模型进行预测需要什么代码?

Thanks in advance. In this post, I asked the question of how to choose specific lags in a VAR model. After a quick reply and information of the 'restrict' and 'coef' functions I was able to successfully run a VAR model with the specific lags I wanted. However, what's the code I need to use the restricted VAR model to make forecasts?

我的代码示例如下:

 ##Attempt to Restrict VAR Coefficients
 ##VAR has 5 lags with three variables plus constant and 11 seasonal dummies.

 library("vars")
 var1 <- VAR(DVARmat, p = 5, type ="const", season = 12)
 restrict <- matrix (c(1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0,
                       1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0,
                       1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0),
                     nrow = 3, ncol = 27, byrow = T)
 var1_restrict <- coef(restrict(var1, method ="man", resmat = restrict))
 var1_restrict

我知道普通 VAR 之后的预测代码，但似乎无法将受限 VAR 加入其中.再次感谢.

I know the forecast code after a normal VAR, but can't seem to fudge the restricted VAR into it. Thanks again.

推荐答案

生成受限系数矩阵restrict后，可以在restrict(...) 因为 restrict(...) 也返回一个 varest 类的对象:

After generating the restricted coefficient matrix restrict, you can use predict on restrict(...) since restrict(...) also returns an object of class varest :

 ##Attempt to Restrict VAR Coefficients
     ##VAR has 5 lags with three variables plus constant and 11 seasonal dummies.

         library("vars")
         var1 <- VAR(DVARmat, p = 5, type ="const", season = 12)
         restrict <- matrix (c(1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0,
                               1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0,
                               1,0,0,1,0,1,0,0,0,0,0,1,0,1,0,0,0,1,0,0,1,0,1,0,0,0,0),
                             nrow = 3, ncol = 27, byrow = T)
         var1_restrict <- coef(restrict(var1, method ="man", resmat = restrict))
         var1_restrict

         expostrestrict <- predict(restrict(var1, method="man", resmat = restrict), n.ahead = 13, ci=.95)

请注意，restrict(var1, method="man", resmat = restrict) 是一个可以生成的对象，因此如果愿意，他们也可以使用以下内容:

Note that restrict(var1, method="man", resmat = restrict) is an object that could be generated, thus if one prefers they may also use the following :

restrict_var <- restrict(var1, method="man", resmat = restrict)
expostrestrict <- predict(restrict_var, n.ahead = 13, ci=.95)

这篇关于具有特定 VAR 模型的预测滞后的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！