Time Series Modelling and Simulation
Published in Spring Time Series, 2021
Which model should I pick, ARMA(2,1) or ARMA(2,2)?
In my Time Series Modeling and Simulation course (MAST6390), I tackled an assessment that delved deep into the fascinating world of ARMA models. Using R, I embarked on a journey of data exploration and analysis, starting with the simulation of a time series using an ARMA(2,1) model. Then, I donned my detective hat and unraveled the patterns hidden within the data through time series plots, ACF, and PACF analyses.
- My mission next was to capture the essence of the data by fitting two models: ARMA(2,2) and ARMA(2,1). But alas, only one could be crowned the champion. Guided by statistical clues and model interpretability, I meticulously selected the most fitting candidate.
- To ensure its true reign, I subjected it to rigorous diagnostic tests, using the tsdiag command and Ljung-Box test to scrutinize its every residual.