Performance Analysis of Machine Learning Algorithms on Predicting Grades Based on Lifestyle of College Students in Northern Philippines Angelo C. Arguson Jaimark P. Villamayor Maria Nessa P. Christ the King College of Quirino State University Solomon Science and Technology Andres Bonifacio, Diffun, Marinduque State None Quirino 3401 College Philippines Philippines None angelocarguson@gmail.c j3vil1031 @yahoo.com Philippines om marianessa_solomon @yahoo.com Mark Anthony P. Cezar Dionito F. Mangao Jr. Albert A. Vinluan Christ the King College of Cavite State University - Naic New Era University Science and Technology Campus None None None Philippines Philippines Philippines albert vinluan@gmail.c nmarc2001 @gmail.com dionmangao@cvsu-naic.edu.ph om ABSTRACT Machine learning applications are becoming a more dependable solution in understanding and solving educational and administrative problems in higher education. This study examined the prediction of students' academic performance through learning algorithms of selected state universities. The subject of the study was selected from a total of 519 students who are BSIT students. The data was collected through administration of questionnaires per year level. The collected data included 35 variables. A randomly selected 80% of these, 519 observations, are used to train the classifier models. The remaining 20%, 130 observations, are used as the test data. In order to classify academic performance students, eight data mining approaches were applied based on random forest, decision tree (DT), support vector machines (SVM), K-nearest neighbors (KNN). logistic regression, Naive Bayes (NB), stochastic gradient descent (SGD), and perceptron. Although all the classifier models show comparably high classification performances, random forest and decision tree classifiers was the best with respect to accuracy. In addition, an analysis of the variable importance was done