Development of Sanitary Landfill's Groundwater Contamination Detection Model Based on Machine Learning Algorithms

Abstract

This study describes the development of five machine learning models for the detection of groundwater contamination due to leachate leakage in a sanitary landfill. A prototype was constructed using Arduino Uno, Wi-Fi module, pH, electrical conductivity and temperature sensors. This prototype was used to gather data from the groundwater and leachate samples in the sanitary landfill. The sensors that were used in the study was calibrated prior to the actual data gathering in the sanitary landfill. Five machine learning model based on logistic regression, quadratic discriminant analysis, k-nearest neighbour, decision tree and support vector machine algorithm was trained and evaluated. Matlab software was used in this study for the development of each model. The accuracy of each model was then compared which results to a 97.8% accuracy for KNN, 97.7% for SVM and Decision Tree, 93.7% for quadratic discriminant and 92.6% for logistic regression model. Based on the results, KNN, SVM and decision tree based models provide the highest accuracy for the detection of leachate leakage on the groundwater located in a sanitary landfill.