Workshop: Time Series Forecasting and Analysis
- Selected option:
- Public
- Cost per delegate:
- £960.00
- Duration:
- 2 Days
The Time Series Forecasting and Analysis workshop will be delivered by our partner Peter Clarke of Deva Statistical Consulting Limited.
Audience
The workshop is designed for analysts who want to make their own forecasts from past data, or who want to assess the effectiveness of a campaign or other factor on the level of business activity.
Prerequisites
Familiarity with basic statistical concepts, such as hypothesis testing and regression, such as could be obtained by attending Amadeus’ business statistics course.
Objectives
By the end of the workshop, delegates should have a sound grasp of the key concepts behind time series forecasting and analysis, and have the confidence to perform their own analysis of time series data. They will have encountered a variety of models, from simple exponential smoothing to ARIMA and structural models, and be able to appreciate the relative strengths and weaknesses of each. Enterprise Guide, SAS/Graph, SAS/STAT and SAS/ETS will be used for the workshop.
*Important Information - Workshop Style*
This workshop concentrates on practical data analysis. There will be very little in the way of lecturing: delegates will be expected to actively contribute to the exploration and analysis of data. Each session will introduce new methodology and build on the methods presented and explored in previous sessions, with the strengths and weaknesses of each compared to alternatives. The written material will not attempt to be a textbook of times series methodology, nor a dictionary of SAS code. It will record the practical steps to be followed to successfully analyse a set of business problems that span a wide range of analysis and forecasting tasks. It will also contain a roadmap to assist with future modelling tasks.
Topics
lntroduction
- What is a time series?
- Why is time series analysis different to other forms of statistical analysis?
TS1 Forecasting
- Simple series with no trend or seasonality
- Moving averages
- Simple Exponential Smoothing
- Random Walks
- Assessing model fit
TS2 Time Series Regression
- Effect of an intervention on a time series
- Dealing with seasonality
- Accounting for trend
- Autocorrelation and autoregressive errors
- Model assessment
TS3 Forecasting
- Forecasting a series with trend and seasonality
- Advanced smoothing methods
- Random walk with drift
- Stationarity
- ARIMA models
TS4 Time Series Regression
- Incorporating dynamic regressors
- Cointegration
- Cross correlation functions
- Prewhitening
- Model identification
- ARIMAX
- Structural models
TS5 Forecasting
- Forecasting a series with variable trend
- Confidence limits for forecasts
- Series decomposition
- Identifying cycles
- Transforming data
- Outlier detection and other diagnostics
TS6 Time Series Regression
- Distributed lag models of advertising effectiveness
- ARIMAX models with denominator terms
- Transfer functions
- Trigonometric methods for seasonality
- Assessment of channel synergy
Conclusions
- A roadmap to determine which time series method to use
- “Soft forecasting” – incorporating expert knowledge into forecasts
- What is not covered on the workshop – where next for time series analysts?

