Add IRF method for VAR

Ref: #363

NAMESPACE

@@ -1,5 +1,6 @@
 # Generated by roxygen2: do not edit by hand
 
+S3method(IRF,VAR)
 S3method(breusch_godfrey,TSLM)
 S3method(components,ETS)
 S3method(fitted,AR)
@@ -136,5 +137,6 @@ importFrom(stats,residuals)
 importFrom(stats,sd)
 importFrom(stats,time)
 importFrom(stats,ts)
+importFrom(tsibble,as_tsibble)
 importFrom(utils,tail)
 useDynLib(fable, .registration = TRUE)

NEWS.md

@@ -2,7 +2,7 @@
 
 ## New features
 
-* Added generate() method for VAR models
+* Added `generate()` and `IRF()` methods for VAR models.
 
 # fable 0.3.4
 

R/var.R

@@ -429,3 +429,30 @@ generate.VAR <- function(x, new_data, specials, ...){
   new_data[colnames(coef)] <- split(.sim, col(.sim))
   new_data
 }
+
+#' @export
+IRF.VAR <- function(x, new_data, specials, impulse = NULL, ortho = FALSE, ...) {
+  new_data$.innov <- matrix(0, nrow = nrow(new_data), ncol = ncol(x$last_obs),
+                            dimnames = dimnames(x$last_obs))
+  # Zero out end of data
+  x$last_obs[seq_along(x$last_obs)] <- 0
+
+  # Remove regressors
+  n_ar <- x$spec$p*ncol(x$coef)
+  if(nrow(x$coef) > n_ar) {
+    x$coef[seq(n_ar + 1, nrow(x$coef)),] <- 0
+  }
+
+  # Add shocks
+  new_data$.innov[1, impulse] <- 1
+
+  # Orthogonalised shocks
+  if(ortho) {
+    # Use Cholesky decomposition to orthogonalise the shocks / innovations
+    new_data$.innov <- new_data$.innov %*% chol(x$fit$sigma2[[1L]])
+  }
+
+  irf <- generate(x, new_data, specials)
+  irf$.innov <- NULL
+  irf
+}