Description
Presentation of the main forecasting methods necessary for decision making in the presence of uncertainty. General principles of forecasting methods used are outlined.
Students will become familiar with the use of key techniques such as smoothing, regression, time series and neural networks. Methods for model evaluation and selection, as well as methods for estimating forecast errors, are also on the program. The R software will be used.
Themes covered
1. Basic ideas good and bad practices evaluation methods
2. Basic tools in forecasting: Naive predictions; ACF PACF stationarity differentiation; Expert opinion
3. Exponential smoothing: Simple Holt Holt-Winters and state-space models; Double seasonal methods
4. Multiple regression: Durbin-Watson test; Linear regression models with ARMA errors
5. Time Series: ARIMA SARIMA and ARMAX models
6. Artificial neural networks: Structure and estimation
7. Multivariate time series
Important notes
Course in French : MATH 60638