Abstract:Accurate multi-horizon retail forecasts are critical for inventory and promotions. We present a novel study of weekly Walmart sales (45 stores, 2010–2012) using a Temporal Fusion Transformer (TFT) that fuses static store identifiers with time-varying exogenous signals (holidays, CPI, fuel price, temperature). The pipeline produces 1–5-week-ahead probabilistic forecasts via Quantile Loss, yielding calibrated 90% prediction intervals and interpretability through variable-selection networks, static enrichment, and temporal attention. On a fixed 2012 hold-out dataset, TFT achieves an RMSE of $57.9k USD per store-week and an $R^2$ of 0.9875. Across a 5-fold chronological cross-validation, the averages are RMSE = $64.6k U…
Abstract:Accurate multi-horizon retail forecasts are critical for inventory and promotions. We present a novel study of weekly Walmart sales (45 stores, 2010–2012) using a Temporal Fusion Transformer (TFT) that fuses static store identifiers with time-varying exogenous signals (holidays, CPI, fuel price, temperature). The pipeline produces 1–5-week-ahead probabilistic forecasts via Quantile Loss, yielding calibrated 90% prediction intervals and interpretability through variable-selection networks, static enrichment, and temporal attention. On a fixed 2012 hold-out dataset, TFT achieves an RMSE of $57.9k USD per store-week and an $R^2$ of 0.9875. Across a 5-fold chronological cross-validation, the averages are RMSE = $64.6k USD and $R^2$ = 0.9844, outperforming the XGB, CNN, LSTM, and CNN-LSTM baseline models. These results demonstrate practical value for inventory planning and holiday-period optimization, while maintaining model transparency.
| Comments: | 5 pages, 2025 6th International Conference on Data Analytics for Business and Industry (ICDABI) |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); General Economics (econ.GN) |
| Cite as: | arXiv:2511.00552 [cs.LG] |
| (or arXiv:2511.00552v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2511.00552 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Praveen Damacharla [view email] [v1] Sat, 1 Nov 2025 13:34:29 UTC (937 KB)