Comparative Evaluation of Machine Learning Models for Residential PM<sub>1</sub> Prediction in Zagreb (Croatia): Identifying Key Predictors and Indoor/Outdoor Dynamics
Source check: Source checked. Not independent scientific confirmation.
Finding: A Zagreb residential pilot measured paired indoor and outdoor PM1 in 103 samples. Gradient-boosted regression explained about 65% of test-set variation in indoor PM1; outdoor PM1, infiltration and resuspension ranked among leading predictors. Winter outdoor peaks did not coincide with higher indoor medians. This local observational result is not a causal estimate or broadly validated exposure rule.
Technology: Gradient-boosted regression predicted indoor PM1 from paired environmental and dwelling variables; SHAP attributed model influence.
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- Gradient-boosted regression predicted indoor PM1 from paired environmental and dwelling variables; SHAP attributed model influence.
- Test and result
- The study evaluated predictions on a held-out portion of 103 paired residential samples and used SHAP to rank predictors.
- Limitation / still unknown
- Small local pilot; model associations do not establish causal effects and may not generalize beyond sampled residences.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://europepmc.org/articles/PMC13120560
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 22 Nov 2026. Review scope: Primary full text, including Methods and Results; publication date and first result-bearing version checked..
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 2/5 · Validation 2/5 · Novelty 3/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.