2026-10-11T01:20:58Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33315021 2026-10-01T21:18:37Z category_28864 item_type_3 month_year_10_2026

RegionFormer / CalST-Bench — Data and Code Deposit Anonymous Anonymous (4854526) Artificial intelligence not elsewhere classified RegionFormer CalST-Bench spatio-temporal forecasting time series forecasting geospatial artificial intelligence reproducibility <p dir="ltr">An anonymised data and code deposit accompanying the manuscript “RegionFormer: Latent-Region Attention for Spatio-Temporal Forecasting,” submitted to the International Journal of Geographical Information Science. The deposit includes the ten-channel CalST-Bench benchmark, model and evaluation code, trained checkpoints, result records, and scripts needed to reproduce the manuscript’s figures, tables, and reported metrics. See README.md and REPRODUCE.md inside the archive for data provenance, licensing notes, setup instructions, and per-finding reproduction commands. The raw SafeGraph point-of-interest panel is not redistributed; the archive contains only the derived, NAICS-aggregated, tract-level visit series used by the mobility channels.</p> 2026-10-01T21:18:37Z Dataset Dataset 10.6084/m9.figshare.33315021.v3 https://figshare.com/articles/dataset/RegionFormer_CalST-Bench_Data_and_Code_Deposit/33315021 CC BY 4.0