LCF attends ISIR 2016 conference

Image by Anita Benesch

30 August 2016

Several members of the Lancaster Centre for Forecasting presented their recent research studies at the 19th International Symposium on Inventories in Budapest. ISIR is the only international scholarly forum focusing directly on inventories.

John Boylan gave a presentation on 'Non-Parametric Estimation for Intermittent Demands'. Bias and variance properties of bootstrapping methods were studied in this research. Their performance has also been compared with the performance of other non-parametric methods.

Nikolaos Kourentzes presented a research on topic of 'Demand forecasting by temporal aggregation: using optimal or multiple aggregation levels?'. He discussed existing approaches to forecasting using different temporal aggregation levels and argued that use of multiple aggregation levels, instead of selecting an optimal one, increases forecasting accuracy.

Nikolaos Kourentzes also provided research in collaboration with Ivan Svetunkov on the topic of 'Asymmetric prediction intervals using half moment of distribution'. A new method of prediction intervals construction in cases of skewed distributions has been proposed and it was shown that there are benefits of this method in comparison with other existing ones.

Patrick Saoud delivered a presentation on the topic of 'Estimating Demand Uncertainty Over Multi-Period Lead Times'. He studied several existing methods of estimating Order-Up-To level, showed their disadvantages and proposed a new method,that takes into account covariances between multiple steps ahead forecast errors. He then showed using simulations that conventional method performs worse than the proposed.

Devon Barrow presented 'Combining and Pooling Forecasts Based on Selection Criteria'. He discussed existing model selection and combination methods and showed that combinations perform on average better than selection. He then compared performance of combinations based on information criteria and cross validation using a pooling technique. The results show that pooling improves forecasting accuracy.

Yves Sagaert gave a presentation on 'Incorporating macro-economic leading indicators in inventory management'. He showed that using LASSO for hierarchies improves long term forecasts accuracy, while in short term it leads to service level and inventory improvement.

Finally Professor John Boylan, director of LCF, has been elected as vice-president and president-elect of the International Society for Inventory Research. He will take up his office as president in 2018, succeeding the current president, Professor Ruud Teunter, who is a former member of LCF.

Image courtesy of International Society for Inventory Research