Intelligent Waste Management and Recommendation System (IWMRS)

Cities in developing countries are growing fast, and their solid waste systems are struggling to keep up. Collection routes are often fixed in advance, records are scattered across departments, citizen complaints take too long to resolve, and planners have little forward-looking evidence to work with. In this paper we describe the Intelligent Waste Management and Recommendation System (IWMRS), a software and machine learning framework that helps municipal staff handle complaints and plan waste operations. The system is built from four connected parts: a React.js web portal with map-based reporting, a Node.js/Express gateway, a Python FastAPI service that hosts the trained models, and a MySQL database normalized to third normal form. Three analytical components sit on top of this stack. The first is a complaint priority classifier that sorts grievances into Low, Medium, High and Critical levels using only information available at the time of filing, so that no resolution details leak into training. The second is a waste generation regressor that estimates daily tonnage for 34 cities using population density, efficiency scores, awareness campaigns and recycling rates, and is tested on data from a later year than it was trained on. The third is a rule-based recommendation engine that compares each city, material, landfill and ward against benchmarks computed from the data itself. We tested the system on 5,000 complaint records and waste records covering 2019 to 2023. Service endpoints respond well under 100 ms. The predictive models, however, carry very little signal on this dataset: the best priority classifier reaches 27.2% accuracy against a 26.5% majority-class baseline, and the waste regressor gives an R² of 0.018 on the standard test split and −0.065 when 2023 is held out. We report these results as they are and discuss why. We present the system as a reusable engineering blueprint for municipal waste administration.