Bayesian Intermittent Demand Forecasting for Large Inventories

Authors: Matthias W. Seeger, David Salinas, Valentin Flunkert

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In a study on large real-world sales datasets, our method outperforms competing approaches on fast and medium moving items.
Researcher Affiliation Industry Matthias Seeger, David Salinas, Valentin Flunkert Amazon Development Center Germany Krausenstrasse 38 10115 Berlin matthis@amazon.de, dsalina@amazon.de, flunkert@amazon.de
Pseudocode No The paper describes algorithmic steps but does not include structured pseudocode or an algorithm block.
Open Source Code No The paper does not provide concrete access to its source code. It mentions 'the lack of public code for [6] precludes a direct comparison', implying their own code is not publicly available.
Open Datasets Yes Parts contains monthly demand of spare parts at a US automobile company, is publicly available, and was previously used in [10, 15, 6].
Dataset Splits Yes We tune7 such parameters on random 10% of the data, evaluating test results on the remaining 90%.
Hardware Specification Yes Our experimental cluster consists of about 150 nodes, with Intel Xeon E5-2670 CPUs (4 cores) and 30GB RAM.
Software Dependencies No The paper mentions software like 'Apache Spark', 'R package [9]', 'L-BFGS', and 'Kalman smoothing', but does not specify version numbers for any of these components, which are necessary for reproducibility.
Experiment Setup Yes We employ quadratic regularization for all methods except ETS (see Section 3.2). Hyperparameters consist of regularization constants ρj and centers θj (full details are given in the supplemental report). We tune7 such parameters on random 10% of the data, evaluating test results on the remaining 90%.