Follow on to Forecasting : ARIMA modelling for forecasting
This course is a natural follow on for delegates who have completed the Foundation course in forecasting and now wish to develop their toolbox further with Autoregressive Integrated Moving Average (ARIMA).
Description
The course is designed for those who have a basic knowledge of forecasting (e.g., smoothing and regression) who wish to take both diagnostic and time series forecasting approaches to the next level.
Learning objectives
- The underlying assumptions of ARIMA models
- The main characteristics of autoregressive and moving average models as used in ARIMA
- The process of the Box-Jenkins methodology for identifying suitable ARIMA models
- How to apply the Box Jenkins methodology to both non-seasonal and seasonal data sets
- To understand when ARIMA models may be appropriate in practice
Topics
- Underlying principles and useful tools:
- Stationarity and white noise
- Autocorrelation
- Differencing
- The suite of models used in ARIMA:
- Models for stationary processes:
- Autoregressive: AR(p)
- Moving average: MA(q)
- Mixed autoregressive moving average: ARMA(p,q)
- Models for non-stationary processes
- Autoregressive integrated moving average: ARIMA(p,d,q)
- Seasonal ARIMA(p,d,q)(P,D,Q)s models
- Box Jenkins Methodology
- Identifying, fitting, and checking models.
Audience
The course is designed for those who undertake time series forecasting.
Ideally, delegates will have a working knowledge of:
- Correlation, confidence intervals, sampling
- Trend and seasonality
- Error measurement including backfit vs forecast error
- Short/medium term forecasting methods e.g., smoothing and regression
Course format
- PowerPoint presentation to introduce the topics
- Practical sessions using Minitab software and Microsoft Excel with the analysis tool pack
- Group discussion/work to explore the topics in more detail
- Bring questions from your own work to embed your learning
- Supporting resource pack available to use following the course
Related courses
- Foundations of OR: Statistical Methods in OR: Multivariate Statistics
- Foundations of OR: Optimisation and (Meta-) Heuristics
- Foundations of OR: Data Envelopment Analysis