This was the final group project of the Time Series course I took during Winter 2020 at McMaster University, in which we modelled the Lake Erie water levels from 1921 to 1970 (public dataset) in R. Project received 95% grade.
Observing the seasonal trends and non-seasonal autocorrelation at many lags in the data, it was deemed suitable to fit the data with Seasonal ARIMA (SARIMA) models.
Two SARIMA models were selected to proceed to the prediction stage for their lower AICs, as well as the number of parameters in the model, with respect to the Principle of Parsimony.
Both models were used to perform a 10-year prediction; both predicted similarly and reasonably in the end.
