Multi-Source County-Year Dataset for Tertiary Industry Added Value Reconstruction in Shanxi Province, China, 2012–2022

This dataset provides the processed county-year panel, metadata, and reproducible files used for reconstructing county-level tertiary industry added value in Shanxi Province, China, during 2012–2022. The dataset integrates multi-source remote sensing, open geospatial, and socioeconomic information within a unified county-year framework. The repository includes processed predictor tables derived from remote sensing, gridded environmental background data, open geospatial data, and socioeconomic statistics, together with the target variable used for model training and evaluation. It is designed to support reconstruction, feature selection, interpretability analysis, and spatial consistency assessment at the county level. Main contents of the dataset include: processed county-year panel data for 2012–2022; feature tables for the full feature set and selected feature subsets; variable descriptions and metadata; reproducible code and supporting files for model fitting, SHAP analysis, and spatial evaluation. The original raw data sources include Landsat, MODIS, MAIAC, ERA5-Land, SRTM, OpenStreetMap, POI datasets, county and city statistical yearbooks, official CO₂ emission inventories, Intangible Cultural Heritage registries, and A-level tourist attraction data. This repository shares the processed analytical dataset rather than redistributing all raw third-party source data.