Intermittent Demand Forecasting · There has been a rich stream of research on intermittent demand forecasting over the last decade, ... performance measurement and organisational

Forecasting Workshop

Tel: +44 (0) 1524 593867| Fax: +44 (0) 1524 844885 | email: [email protected]

Intermittent Demand Forecasting

The Work Foundation | LONDON February 27, 2015 - 13:00-17:15

with presentations from:

Tel: +44 (0) 1524 593867 | Fax: +44 (0) 1524 844885 | email: [email protected]

Open Workshop | Intermittent Demand Forecasting

About the workshop

Welcome to the half-day workshop on Intermittent Demand Forecasting. Intermittent demand is prevalent is many industry sectors and arises whenever there are zero observations in a time series. These series are common in the military, automotive and aerospace sectors but are also common in any in-dustry with an aftermarket for service parts. Now that customer level data is becoming increasingly easy to obtain, intermittent demand patterns are arising in sectors such as FMCG when analysing highly disaggregated data. There has been a rich stream of research on intermittent demand forecasting over the last decade, leading to new approaches and new techniques to deal with this type of demand. At the same time, practitioners have been working on finding better ways of dealing with some of the practical issues which have been neglected by academic research-ers. This workshop aims to:

Give an overview of advances in the field, focussing on those which may be implemented most readily.

Explain practical approaches to assess which intermittent demand items should be stocked at all.

Highlight the pitfalls of using standard accuracy measures and identify alternative measure which are better suited to intermittent demand.

Review the freeware that is now available from the Lancaster Centre for Forecasting, which will enable you to implement the new methods outlined in the workshop.

There will be industry talks from Euro Car Parts and Satori Partners and the Forecasting Centre, who will share their insights of working with intermittent demand data and explain how they have overcome some of the challenges presented by this type of demand pattern. The workshop will conclude with a panel discussion, allowing all delegates to contribute to the debate and to pose questions to the panel. We look forward to welcoming you in London.

Recent advances in Intermittent Demand Forecasting by Prof. John Boylan (Lancaster Centre for Forecasting) What methods are available for intermittent demand forecasting? In this talk, the pros and cons of Croston’s method (the ‘industry standard’) will be examined, together with adaptations, now available as freeware. Recent work on ‘temporal aggregation’ will be reviewed. This approach, which is simple to apply and easy to explain, is based on aggre-gating the history of demand into natural time blocks. The talk will conclude with remarks on assessing the performance of intermittent demand forecasts, a theme which will be taken up in more depth by the third talk in this workshop.

Practical Issues in Intermittent Demand Forecasting

by Roy Johnston (Formerly Euro Car Parts Ltd) What should be the breadth of range held by branches within a supply chain, and how should this be related to the quan-tities of stock to be held at stocking points? In this talk, a case-study will be presented on how these questions were ad-dressed at Euro Car Parts. The approach was based on balancing profit and stock-holding costs. An appropriate stock/non-stock boundary was established, and this approach was extended to rules about stocking two or more units of an SKU. The talk will conclude with a review of the financial impact of implementing the new rules.

Assessing the Performance of Intermittent Demand Forecasts

by Steve Morlidge (Satori Partners)

How can we compare the accuracy of different forecasting methods for intermittent demand, and how should these measures relate to inventory performance? In this talk, some of the pitfalls of using the most commonly applied forecast error measures will be outlined. Although these measures may work well for fast-moving demand patterns, the same is not true for intermittent demand. A new approach for measuring performance will be outlined, which is easy to apply in practice and which has links to measures which are appropriate for inventory management.

Panel Discussion: Making Better Forecasts in Practice The panel will consist of John Boylan (Lancaster Centre for Forecasting (LCF), Roy Johnston (Euro Car Parts), Steve Morlidge (Satori) and Nikoloas Kourentzes (LCF), discussing a range of topics relating to the practice of intermittent de-mand forecasting. Delegates will have ample opportunity to ask questions of the panel and to put forward their own views. The panel discussion will be chaired by Prof Robert Fildes (LCF).

Programme 13:00-13:45

13:45-14:30

14:45-15:30

16:15-17:00

17:15 onwards

coffee break

Freeware for Intermittent Demand Forecasting By Nikolaos Kourentzes (Lancaster Centre for Forecasting)

What software is available to support intermittent demand forecasting? In this talk, there will be brief review of the facili-ties available in major software packages such as SAP APO and Forecast Pro, as well as more specialist packages such as SmartForecasts. Criteria to assess software packages will be outlined, and the limitations of current software will be sum-marised. Then, the session will move to a presentation on “Intermittent demand forecasting package for R”, which is freeware developed by Dr Kourentzes and colleagues at the Lancaster Centre for Forecasting.

15:30-16:15

Networking Reception Location: Adam & Eve, 81 Petty France (corner of Palmer Street), London SW1H 9EX

Tel: +44 (0) 1524 593867 | Fax: +44 (0) 1524 844885 | email: [email protected]

Open Workshop | Intermittent Demand Forecasting

Location

Workshop location The Work Foundation (21 Palmer Street, LONDON,

SW1H 0AD)

Easily accessible, just 3 minutes walk from the St. James's Park Underground Station

Interested?

John Boylan Lancaster Centre for Forecasting John has recently taken up his post as Professor of Business Analytics at Lancaster, but has been active in fore-casting research and practice for over 25 years. He has a background in the aerospace and automotive indus-tries, where he first encountered the challenges of intermittent demand forecasting. He was one of those, in the 1990s, who helped to revive interest in this field. He has published extensively on the subject of inter-mittent demand, and his name is associated with one of the standard benchmark methods for forecasting these demand patterns. He has worked with software companies on the implementation of new forecasting methods and has given many talks to industry practitioners on this subject.

Roy Johnston Formerly Euro Car Parts Ltd Roy has had extensive experience of the implementation of forecasting and stock control systems over many years. He has worked with builders merchants, electrical wholesalers, a cider manufacturer and some national brewers, chemical companies, and car parts distributors, using a wide variety of forecasting methods but al-ways relating these to the production or stocking decisions. Parts with intermittent demand can frequently comprise the majority of a stockist range and to produce decision rules for their stock management places specific requirements on a forecasting system, including forecasting and characterising the size of orders. Most recently, he has looked into the development of rules on stock ranges, which depend on a forecast of the number of orders received for each Stock Keeping Unit at each stocking unit. This work led to the award of the Goodeve Medal, one of the highest accolades of the Operational Research Society.

Steve Morlidge Satori Partners Steve has 30 years of practical experience, designing, and running performance management systems in Unilever. He is a former Chairman of the European Beyond Budgeting Round Table and now works helping companies break out of traditional, top down ‘command and control’ management practice with special emphasis on the role of forecasting at the interface between finance and supply chain. He has recently published ‘Future Ready: How to Master Business Forecasting’ (John Wiley, 2010), is a Visiting Fellow at Cranfield University, has a PhD in Managerial Cybernetics at Hull Business School and is on the editorial board of Foresight to which he regularly contributes. He is also cofounder of CatchBull, a supplier of forecasting performance management software.

Nikolaos Kourentzes Lancaster Centre for Forecasting Nikos is a lecturer at the department of Management Science at Lancaster University. He has been active in forecasting research and consultancy for the past ten years. One of his research interests is intermittent de-mand, where he has focused mostly on how to improve the robustness and automation of forecasting meth-ods.

About the Speakers

If you are interested in events like this one, please register to our mailing list through our website: www.forecasting-centre.com

Tel: +44 (0) 1524 593867| Fax: +44 (0) 1524 844885 | email: [email protected]

Open Workshop | Intermittent Demand Forecasting

LCF is part of the renowned Lancaster University Management School (LUMS), one of the UK's top busi-ness schools with an outstanding research and teaching record. LUMS is among the world's most highly ranked business schools, consistently listed in the UK's top five for its academic degrees, and one of the few business schools in the world to hold triple accreditation (AACSB, EQUIS and AMBA). Lancaster University Management School won the inaugural Business School of the Year award in the Times Higher Education awards 2012.

The Centre is based within the distinguished Department of Management Science, the 1st to be founded in the UK with a 50-year tradition dating back to Gwilym Jenkins (of Box-Jenkins/ARIMA fame). Today, with over 40 faculty one of the largest of its kind, it has a worldwide reputation as a centre of excellence in OR/MS.

In teaching at LUMS, members of the LCF play a vital role in educating students and executives in Operational Research and Operations Management in forecasting methods, software packages and processes, and in supervising 20 Master's and PhD projects in forecasting with industry every summer. As a result, engaging with LCF gives companies access to some of the leading researchers in the UK, cost efficient projects with MSc and PhD students, and access to recruit some of the best analytical talent in the UK—firmly educated in forecasting, data mining and predictive analytics.

The Lancaster Centre for Forecasting (LCF) leads the field in applied forecasting research, consultancy and training in Europe. A not-for-profit-organisation, LCF engages with companies in projects to facilitate knowledge-transfer between academia and business, and to establish and disseminate best practices.

LCF offers over 20 years of forecasting expertise. Founded in 1990 by Prof. Robert Fildes, past president of the International Institute of Forecasting (IIF), co-founder of the International Journal of Forecasting and the Journal of Forecasting, it was the first centre to focus on forecasting research and practice. Today, with 20 members (5 senior members supported by 2 post-doctorate research associates, 2 software developers and 11 PhD students), LCF hosts the largest think-tank dedicated to forecasting worldwide.

The Centre offers services on the complete range of predictive analytics, from demand planning to market modelling, from statistical methods to artificial intelligence, and from public sector to corporate solutions in telecommunications, fast moving consumer goods, insurances and manufacturing. Members regularly support companies in their use of forecasting methods, auditing forecasting processes (including performance measurement and organisational setup), and selecting and tuning forecasting software systems.

In addition, the Centre offers a range of training courses in forecasting and demand planning, introducing the fundamentals of forecasting methods (e.g. on Exponential Smoothing, Causal Regression) or advanced topics (e.g. on ARIMA models, Neural Networks). To better address a company’s particular situation and learning needs, LCF also offers tailor made courses held in-house. Contact us to find the most suitable form of training to tackle your current organisational challenges!

Some of the images in this brochure were purchased from the: © iStockphoto.com

About us