Added solution to A4

assignments/A4.R

@@ -0,0 +1,35 @@
+generate_arma <- function(n = 100, c = 0, phi = NULL, theta = NULL, sigma = 1) {
+  # Model parameters
+  if (is.null(phi)) phi <- numeric()
+  if (is.null(theta)) theta <- numeric()
+  p <- length(phi)
+  q <- length(theta)
+  if (any(abs(polyroot(c(1, -phi))) <= 1)) {
+    stop("Model is not stationary")
+  }
+  if (any(abs(polyroot(c(1, theta))) <= 1)) {
+    stop("Model is not invertible")
+  }
+
+  # Series length
+  if (n <= 0) {
+    stop("n must be positive")
+  }
+  # Add 50 for burn-in period
+  nplus <- n + max(50, p, q)
+
+  # Generate errors
+  error <- rnorm(nplus, mean = 0, sd = sigma)
+
+  # Set up vector for the response with initial values
+  # equal to the mean of the process for faster burn-in.
+  y <- rep(c / (1 - sum(phi)), nplus)
+
+  # Generate remaining observations
+  for (i in max(p, q) + seq_len(nplus - max(p, q))) {
+    y[i] <- c + sum(phi * y[seq(i - 1, by = -1, length.out = p)]) + sum(theta * error[seq(i - 1, by = -1, length.out = q)]) + error[i]
+  }
+
+  # Return last n observations
+  tail(y, n)
+}

assignments/A4.qmd

@@ -68,3 +68,9 @@ Please submit your solution as a .R file.
 source(here::here("course_info.R"))
 submit(schedule, "Assignment 4")
 ```
+
+---
+
+## Solution
+
+[R code](A4.R)