Employing Machine Learning Models in Prediction of Harmful Gases from Agri-Waste

Climate destruction is one of the major concerns for the government today. Out of many, one crucial factor of atmospheric rapidly changing condition is the harmful gases released through material waste. Agriculture waste also comes in this category. Prediction of the amount of gases to be produced plays a very important role to control the production of such gases. Present study presents a novel machine learning-based prediction and classifier models to predict the production of harmful gases from agri-waste. After implementing feature optimization and selection over the collected dataset, the random forest algorithm achieved the 100% accuracy for classification followed by logistic regression (97.91%), AdaBoost (97.89%), and decision tree (97.84%). In regression models, SVM gave the highest prediction accuracy of 98%. This work also focuses upon the exploratory data analysis of various features available in the dataset which is a helpful tool to provide a guidance for reducing gas emissions from agriculture waste. The results show that EDA has proven to be a powerful tool for performance improvement of machine learning algorithms for the stated problem.