Comparison of Machine Learning Algorithms for the Classification task

Abstract

This study presents a comprehensive evaluation of multiple classifiers using precision metrics in the context of a car evaluation dataset obtained from the UCI Machine Learning Repository. The classifiers under consideration include Decision Tree, Naive Bayes, and Artificial Neural Network (ANN). Precision values, representing the accuracy of positive predictions among all instances predicted as positive, are utilized as a key performance indicator. The Decision Tree classifier demonstrates exceptional precision across multiple classes, particularly excelling in the ‘unacc’ category. In contrast, the Naive Bayes classifier exhibits lower precision, suggesting potential challenges in accurately identifying positive instances. The ANN classifier falls between the other models, showcasing a balanced precision performance. Furthermore, the precision values are analyzed in conjunction with other performance metrics such as recall, F1-score, and accuracy. The study goes beyond individual precision values, considering weighted average precision, which accounts for class imbalances, and macro average precision, treating each class equally. The results provide nuanced insights into the strengths and weaknesses of each classifier, offering a valuable guide for selecting an appropriate model in the context of the car evaluation dataset. The findings contribute to the broader understanding of classifier performance evaluation, emphasizing the importance of precision metrics in real-world applications.