Assignment 3

This assignment will use the same data that you will use in the retail project later in the semester. Each student will use a different time series, selected using their student ID number as follows.

library(fpp3)
get_my_data <- function(student_id) {
  set.seed(student_id)
  all_data <- readr::read_rds("https://bit.ly/monashretaildata")
  while(TRUE) {
    retail <- filter(all_data, `Series ID` == sample(`Series ID`, 1))
    if(!any(is.na(fill_gaps(retail)$Turnover))) return(retail)
  }
}
# Replace the argument with your student ID
retail <- get_my_data(12345678)

Use a training set up to and including 2018.

• What transformations (Box-Cox and/or differencing) would be required to make the data stationary? You should use a unit-root test as part of the discussion.
• Use a plot of the ACF and PACF of the (possibly differenced) data to determine two plausible ARIMA models for this data set.
• Fit both models, along with an automatically chosen model, and produce forecasts for 2019–2022.
• Which model is best based on AIC? Which model is best based on the test set RMSE? Which do you think is best to use for future forecasts? Why?
• Check the residuals from your preferred model, using an ACF plot and a Ljung-Box test. Do the residuals appear to be white noise?

Submit a Quarto (qmd) file which carries out the above analysis. You need to submit one file which implements all steps above. You may use this file as a starting point.

---

Due: 16 May 2025
  Submit