ARBA MINCH UNIVERSITY ARBA MINCH INSTITUTE OF TECHNOLOGY FACULTY OF COMPUTING AND SOFTWARE ENGINEERING

Abstract

Ethiopia possesses the largest livestock population in Africa. Among this, as Central Statistical Agency reported, the estimated cattle population is more than 60.39 million in 2018. Trypanosomiasis is one of the most crucial protozoan diseases of cattle which affects the health of animals. Computer-based methods are increasingly used to improve the quality of medical services. Machine Learning tools are very effective for the analysis of medical data to solve diagnostic problems. Machine learning algorithms are capable of managing and combining huge number of data from dissimilar resources to give optimal accuracy result for a particular problem, like Trypanosomiasis. The overall prevalence of trypanosomiasis revealed that 11.2% in South region of Ethiopia in 2019 and according to a meta-analysis done in 2016, the pooled estimate of this disease in Ethiopia was 8.12%. This disease result in morbidity and mortality with serious socioeconomic consequences. The objective of this study is to develop a model to predict species of Trypanosomiasis using Machine Learning. The study compared three Machine learning tools (Rapid Miner, Orange and Matlab) using six Algorithms (K-Nearest Neighbor, Artificial Neural Network, Support Vector Machine, Logistic Regression, Random Forest and Decision Tree) by four performance measures (Accuracy, Sensitivity, specificity, and Receiver Operating Characteristics) with cross validation test. The study used 5262 datasets and 11 attributes collected from Arba Minch, Jinka, Sawula and Wolayita. This study followed applied experimental design through mixed approach (Qualitative and Quantitative). It is found that Decision tree had better accuracy of 92.66% using Rapid miner with 10-fold cross validation. Artificial Neural Network using Matlab was the most sensitive with 92.5%. Rapid miner’s K-Nearest Neighbors was the most specific model with 93.98%. Therefore, based on the data analysis and the extracted performance measures results, it is concluded that decision tree was the best model for Trypanosomiasis species prediction and Rapid miner is the best machine learning tool.