Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{209668,
        author = {Dr MK Jayanthi Kannan and SHIVA KUMAR},
        title = {Explainable Machine Learning for Air Pollution Level Prediction and Contributing Factor Analysis},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {13},
        number = {5},
        pages = {4180-4189},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=209668},
        abstract = {Air pollution is a major environmental concern, and short-term forecasting of fine particulate matter (PM2.5) supports monitoring, public awareness, and informed decision-making. The “AirInsight” platform is designed as an explainable PM2.5 prediction system that applies machine-learning regression to historical air-quality observations from the Anand Vihar (Delhi, DPCC) monitoring site and forecasts the PM2.5 concentration one hour ahead. With features including data cleaning and timestamp auditing, a leakage-free one-hour-ahead target, chronological model evaluation, feature-contribution explanations, and an interactive dashboard, AirInsight aims to make air-quality forecasts transparent and reproducible. Linear Regression and Random Forest are compared against a persistence baseline using MAE, RMSE, and R2. On the chronological test set (2,797 rows), Linear Regression achieved MAE 15.477, RMSE 21.539 and R2 = 0.7116; Random Forest achieved MAE 15.734, RMSE 22.507, and R2 = 0.6851, while the persistence baseline achieved MAE 14.583, RMSE 21.342, and R2 = 0.7169. The persistence baseline therefore matched or beat both trained models, showing that strong autocorrelation limits the value of current-hour features and motivating lagged features. Feature contributions describe model behaviour and are not evidence of emission sources.},
        keywords = {PM2.5, Air-Quality Prediction, Machine Learning, Explainable AI, SHAP, Linear Regression, Random Forest, Persistence Baseline, Dashboard.},
        month = {October},
        }