Loading
Life sciences · Preprint
arXiv · September 4, 2026
Early or partial results. Treat as a signal, not a conclusion.
This preprint describes a counterfactual interpretability method for deep sales forecasting models, evaluated for faithfulness using deletion and insertion protocols on a large proprietary retail dataset. The method produces statistically significant effects (deletion gap 0.22, p<0.001; insertion gap 0.27, p<0.01) and avoids allocation artifacts seen in SHAP-style approaches, but is limited to one dataset and one application domain, with acknowledged heterogeneous utility across different series.
Computational methods study; post-hoc interpretability evaluation on a trained multi-series WaveNet model. Corporacion Favorita grocery sales data: 174,685 product series tracked over 1,688 days. No explicit inclusion/exclusion criteria reported. Single-source proprietary dataset.. Intervention: Post-hoc counterfactual interpretability layer generating decomposed attributions of forecasts into contributions summing to predicted value. Compared with: SHAP-style additive attribution methods (noted to exhibit allocation artifacts); no direct head-to-head comparison reported in source.. n = 174,685. Corporacion Favorita (Ecuador-based grocery chain implied by dataset name); no explicit geographic descriptor given..
Deletion gap 0.22 (p<0.001) and insertion gap 0.27 (p<0.01) indicate statistically significant faithfulness of attributions across both tests Faithfulness effect robust across five background-sampling seeds Promotion signal reliance heterogeneous across series with median ratio approximately 1.0; roughly 20% of series show strong effect
Safety was not reported in the material analysed. Check the source before drawing any conclusion about harm.
The source did not state who this applies to in practice.
Methodological study on interpretability of a forecasting model with rigorous faithfulness testing, but no clinical outcomes, no peer review, and limited to a single proprietary dataset.
As stated by the source record.
Quoted from the source exactly as published.
Graded across the dimensions that decide whether you should act, each from what the source actually supports. There is no single score, and where a dimension was not assessed it says so.
Deep models for sales forecasting, such as WaveNet-style dilated convolutional networks, are accurate but opaque: when a single model predicts sales for one of many series, it offers no account of why. We add a post-hoc, architecture-agnostic counterfactual interpretability layer to a multi-series WaveNet forecaster trained on the full Corporacion Favorita grocery dataset (174,685 series over 1,688 days). The method decomposes each forecast into contributions that sum exactly to the predicted value, avoiding the allocation artifacts we observed with additive SHAP-style attribution. We evaluate faithfulness with a deletion/insertion protocol and find a statistically significant effect on both tests (deletion gap 0.22, p<0.001; insertion gap 0.27, p<0.01; robust across five background-sampling seeds), establishing that the attributions reflect genuine model behavior rather than plausible-looking artifacts. We then characterize, honestly, where attribution is and is not informative: reliance on the promotion signal is heterogeneous across series (median ratio approximately 1.0, with roughly 20% of series showing a strong effect), and the model captures the shape of the weekly sales cycle (day-of-week r=0.78) while systematically under-predicting its amplitude. Our contribution is not improved accuracy but an interpretability layer with a rigorous faithfulness evaluation and a candid account of its limits.
Taken from the source record, never inferred. Follow any of these and new work involving them reaches your briefing.