The Intermittent Demand Forecasting event took place at the Work Foundation, where around 35 people from a wide range of sectors gathered to hear about new methods and software to help in this area.

The opening talk was given by , who recently joined Lancaster University as Professor of Business Analytics. John explained that research development had been very slow up until around the year 2000. In the last 15 years, however, research had accelerated with new methods being introduced, many of which had been validated on real data. Unfortunately, he said, there was still a lag between research developments and software implementation.

Next, Roy Johnston presented a case study of intermittent demand forecasting at Euro Car Parts (ECP). ECP have many intermittent demand items, which may be stocked at numerous locations. Roy explained how the development of decision rules had been introduced in the company, helping them decide whether to stock an item and how many to stock. The implementation had been successful, leading to considerable savings in stock investment.

Steve Morlidge, from Satori Partners, gave a presentation on accuracy assessment, based on his consulting experience. He explained how traditional measures, such as Mean Absolute Error, are unsuited to intermittent demand. Steve introduced a new measure, the Bias-Adjusted Mean Absolute Error, showing how it was better suited to intermittent demand and how it linked to measures of effectiveness in inventory management.

The final talk was given by  from the Lancaster Centre for Forecasting. Nikos, together with , had recently developed computer code in R, which is publicly available. This includes approaches such as Croston’s method and the Syntetos-Boylan Approximation. This is a major step forward, as it allows companies to incorporate these methods into their own applications. Click here or email n.kourentzes@lancaster.ac.uk for more information.

The day concluded with a lively question-and-answer session, chaired by Professor . Informal discussion at the close of the day revealed keen interest in working to improve intermittent demand forecasts. Lancaster Centre for Forecasting would be very pleased to hear more from people interested in working with them in this area. Anyone interested should email j.boylan@lancaster.ac.uk