Academy & Industry Research Collaboration Center (AIRCC)

Volume 9, Number 14, November 2019

Dynamic Cluster based Markov Model for Demand Forecasting


Albert F. H. M. Lechner, Steve R. Gunn, School of Electronics and Computer Science University of Southampton, UK


Sales forecasts are essential to every business strategic plans and can both save the business money and increase its competitive advantage. However, many current businesses underestimate the opportunities accurate forecasts provide and rely on judgemental forecasts from experts within the business. Machine learning and statistical forecasting methods are used by both academics and practitioners to increase the accuracy of these forecasting methods and can be further improved by applying the newly developed dynamic cluster based Markov model, presented in this work. This approach gathers global sales pipeline data to build a short-term sales forecast. The prediction of future sales for the next three months is improved over a regular Markov transition model. The new model can support short-term planning, thereby supporting regional and product-specific forecasting to steer business activities to their given targets and remain profitable.


Demand forecast, Time series data, Clustering, Markov model