Sensitivity of Classical Moving-Average Traffic Forecasts

> The datasets used in this study are publicly available benchmark datasets for spatio-temporal> traffic forecasting. Standardized CSV versions of the datasets are provided as Supplementary> Material accompanying this manuscript. The datasets were obtained from their original public> repositories: PeMSD7 and PeMSD7(M) from the STGCN repository> (https://github.com/VeritasYin/STGCN_IJCAI-18); PEMS04 and PEMS08 from the ASTGCN repository> (https://github.com/Davidham3/ASTGCN-2019-mxnet); PEMS03 and PEMS07 from the MAGCN repository> (https://github.com/zqylen/MAGCN); and METR-LA and PEMS-BAY from the DCRNN repository> (https://github.com/liyaguang/DCRNN). All repositories were accessed on 2026-10-11. No> normalization or imputation was applied during format conversion; the original numerical> precision was preserved. The original data are collected by the California Department of> Transportation Performance Measurement System (PeMS) and, for METR-LA, derived from the> Los Angeles County highway network.