TOOL For Predicting Plans and Improving Solid Waste Management by Applying Time Series and AWS

Currently, the constant population growth in Peru leads to the generation of large amounts of solid waste, most of these originate from domestic sources and are collected by municipalities. The inefficiency in solid waste collection plans generates rework and therefore operational overhead. Likewise, it pollutes the environment due to waste accumulated in the streets and dissatisfaction from the neighbors. Therefore, an IoT tool is proposed, which will have a solar panel, sensor, chip inside the containers that served to train the algorithm and predict solid waste collection plans for decision-making of a municipality using the time series model, arima and Random Forest. This tool has 3 stages: analysis, design, and validation. Stage 1 evaluates machine learning models and IoT architectures. Stage 2 describes the design and software architecture of the tool. And, in stage 3, the deployment of IoT devices in a municipality of Lima was simulated and its operation was validated with the solid waste collection area. The result of the implementation of the tool in a sector of Villa El Salvador managed to increase the efficiency of solid waste collection plans from 81 % to 96 % effectiveness in said sector. In addition, this reduced operational costs by approximately 8 % and increased the satisfaction of neighbors by not having solid waste accumulated in the streets.