The International Symposium on Forecasting 2024 took place in Dijon, France, and was well attended by CMAF members.

Sven Crone delivered a talk on "Forecasting with Artificial Neural Networks for Sparse Data - an Empirical Evaluation," discussing neural networks' performance versus statistical and machine learning benchmarks on an FMCG dataset.

Robert Fildes, in collaboration with Paul Goodwin, presented "Judgmental Adjustments of Computer-Based Forecasts: Are They Beneficial? Can They Be Improved?". They analyzed 147,000 forecasts from six studies, finding that adjustments improved accuracy in only half of the stock-keeping units.

Stephan Kolassa's talk, "Beyond Accuracy: What Makes for a Good Forecast?", covered various aspects of accuracy, including interpretability, debuggability, robustness, resource requirements, and computing power, focusing on the impact of forecasts on business value.

Kandrika Pritularga discussed designing regularization estimator to achieve congruency and accuracy in exponential smoothing forecasts in "Forecast Congruency, Accuracy, and Shrinkage Estimators for Exponential Smoothing."

Anna-Lena Sachs presented "Inventory Control with FIFO and LIFO Picking Behavior: The Role of Sustainability Messages and Price Discounts," discussing how these behaviours affect inventory management in retailing and whether retailers' strategies can influence them.

Anna Sroginis presented her joint work with Ivan Svetunkov on "Intermittent or Not? How to Tell the Difference," where they explored how to identify stock-outs and classify demand into intermittent/regular and count/non-count.

Ivan Svetunkov's presentation, "How to Bootstrap Time Series without Attracting the Attention of Statisticians," introduced a modification of the Maximum Entropy Bootstrap and evaluated it alongside the original and STL-based methods on M1, M3, and tourism datasets.

Alisa Yusupova (in joint research with Nicos Pavlidis) delivered a talk on "A Sequential Monte Carlo Approach to Adaptive Exponential Smoothing", suggesting modifications to conventional ETS to detect and capture different types of disruptions.

Ritika Arora presented "Demand Forecasting in the Presence of Supply Chain Disruptions," introducing an ETSX modification that allows the model to adapt during disruptions, better capturing the dynamics and producing more accurate forecasts.

Maddie Smith talked on "Dynamic Forecast Combination Using Point or Density Forecasts", proposing and contrasting two Dynamic Linear Model-based forecast combination methods.

Our honorary researchers, Devon Barrow, Nikos Kourentzes, Oliver Schaer, and Yves Sagaert, also delivered well-received presentations during the conference.

Finally, Ivan Svetunkov and Kandrika Pritularga successfully delivered a workshop titled "Sky is the Limit: Bringing Exponential Smoothing to the Next Level," attended by 27 participants.

Our presentation slides can be found here: