Event Details

This course will improve your skills in business forecasting principles and modelling, leveraging the power of open source software.

We will rely on R statistical computing language to gain a hands-on understanding of business forecasting theory. The course will provide the modern analyst with the know how to provide expert forecasting modelling and advice, both in terms of the technical details, but also in terms of improving the forecasting process within an organisation setting. We will cover state-of-the-art implementations of forecasting model families with proven track record, as well as more advanced forecasting approaches, addressing the complex challenges of business forecasting. Using the long experience of the Centre for Marketing Analytics and Forecasting at Lancaster University, this course will go beyond rehashing forecasting algorithms and code snippets, giving you an in-depth understanding of what happens under the hood, when an approach is fitting to the problem at hand, and what are the limitations of the various methods.

The course is ideal for the business forecaster who wants to gain an understanding of the statistical modelling and how to leverage the power of open source software for their companies, but also for the data scientist who wants to gain an in depth understanding of the business forecasting know-how.

Course topics

Day 1

1. Forecasting principles: what makesf orecasting different from conventional statistical modelling? How does this impact the way we approach statistical forecasting and what does it mean for our business forecasts?
2. Time series exploration and decomposition: what can be captured in a time series? How to identify what is in ourdata?
3. Forecasts evaluation: what is a good forecast? What are useful metrics of forecast performance and how to evaluate our forecasting process?
4. Exponential smoothing family of models: this is the work horse of statistical business forecasting, with compelling evidence from practice and academia of its reliability and performance. We will understand what it can and cannot model, as well as understand how to best implement the models in the data at hand.

Day 2

1. Exponential smoothing with events and promotions: we introduce ways to model special events, bank holidays, promotions, etc. With exponential smoothing, providing the mechanics for advanced statistical business forecasting.
2. The basics of regression: we expand further on causal modelling demonstrating the ease to build regression models that capture effects of exogenous variables and events with ourforecasts.
3. Advanced forecasting methods: we will demonstrate the ease of using more advanced forecasting techniques that buildon the fundamentals of exponential smoothing and regression, without going in the theoretical details. Examples are ARIMA, MAPA and simple Neural Networks, as well as how to increase forecasting quality of life using tools such as the ABC-XYZ analysis.
4. R forecasting resources: we will provide you with an overview of important R packages for business forecasting, giving you examples of the problems one can solve.