Random forests map South African particulate pollution patterns
Source check: Source checked. Not independent scientific confirmation.
Finding: A random-forest analysis combined ground stations, satellite data and environmental predictors to estimate daily particulate pollution around Gauteng during 2014–2018. It mapped persistent hotspots and seasonal differences, with lower estimated concentrations in much of the region after 2016. The comparison does not establish a policy effect.
Technology: Random forest exposure regression
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Learned tree ensembles infer particulate concentrations from satellite, meteorological, land-use and socioeconomic inputs; the field result is the resulting bounded regional exposure pattern.
- Test and result
- The model used 9,853 station-day observations from 20 monitors and 40 predictors. Random ten-fold cross-validation gave R² .80 and RMSE 9.40 µg/m³. Predicted winter concentrations were highest; Johannesburg and surrounding areas showed little clear change between 2014–2015 and 2016–2018 while much of the domain decreased.
- Limitation / still unknown
- Station-day random folds do not demonstrate transfer to unseen monitoring sites or future years. The model underestimates very high concentrations; top-quintile CV R² was .69. Satellite missingness and correction introduce additional uncertainty. A before/after-policy indicator is itself a model predictor, and the comparison has no causal control or counterfactual. Maps are modeled estimates, not complete ground measurements. No individual health benefit is measured.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC8589277/
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: Original ground measurements, integration, random-forest Methods, policy analysis and represented Results actually read..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.