This document presents a study that compares the performance of 10 classification algorithms (Naive Bayes, SMO, KStar, AdaBoostM1, JRip, OneR, PART, J48, LMT, Random Tree) using 3 datasets from the UCI Machine Learning Repository (German credit data, ionosphere data, vote data). The algorithms are tested using the WEKA machine learning tool. The results show that Random Tree and LMT generally have the best predictive performance across the different testing modes and datasets, with Random Tree achieving the highest accuracy on the German credit and vote datasets, and LMT performing best on the ionosphere data.