PMBoost v1.0 Primary Datasets

### OverviewThis repository contains the primary dataset supporting the manuscript on PMboost v1.0. It includes processed model predictions, ground-truth observations for holdout test results and operational test outputs, and output data for hourly spatial plots. --- ### Dataset Contents & Structure 1. Holdout Validation Results ("PMBoost_2016_2020_holdout_test_results.pickle") - Outputs from holdout test for 2016 to 20220 - Columns: UTC_date, PM25_pred, PM25_obs, SSL 2. Operational Test Results ("PMBoost_2021_operational_test_results.pickle") - Outputs from the operational evaluation period (2021), reflecting real-world near-real-time prediction performance. - Columns: UTC_date, PM25_pred, PM25_obs 3. Example Hourly Spatial Plots ("PMBoost_{date}_xgb.csv") - Inputs and outputs for hourly spatial plots for high PM episodes - Figure 5: 2021021206~2021021506 - Figure 6: 2021031004~2021031304 - Figure 7: 2021032704~2021033104 PMBoost ### Units & Technical Notes- **PM2.5 Concentrations:** µg/m³- **Temporal Coverage:** 2016 – 2021 --- ### Usage & CitationPlease cite the accompanying manuscript and this dataset repository when using this data.