Time Series and Forecasting
University of am I here?
▪ Next year in Semester 1, I will be teaching Time Series and Forecasting (QBUS3850).
▪ For the last four weeks you have been learning about forecasting.
▪ In QBUS3850 we extend your knowledge in forecasting theory and practice.
A taste of things to come
▪ Seasonal ARIMA
▪ Vector Autoregressions
▪ Model Combination
▪ Value at Risk Forecasting
Retail Data
ARIMA (1,1,1)
Seasonal ARIMA
Vector Autoregression (VAR)
▪ Often forecast multiple variables that are all part of one system.
▪ Particularly popular in macroeconomics but used elsewhere
▪ Can get better forecasts but can also make scenario based projections.
▪ Following example come from Canadian Labor market data
Impulse response functions
Wisdom of the Crowds
The idea of combining forecasts from different models is a very powerful technique.
Financial returns
▪ While forecasting the mean is hard, forecasting the variance is easier.
▪ This can be used to forecast quantiles.
Forecasting variances and quantiles
▪ My research involves forecasting problems ▪ Energy
▪ Macroeconomics ▪ Online retail
▪ Insurance
▪ I am a director of the International Institute of Forecasters
▪ I have consulted for the International Monetary Fund.
▪ You can find my details on my personal website.