<![CDATA[A student's ability to complete courses is influenced by various factors, including academic and non-academic aspects. Understanding the factors that influence it is very important to know in order to anticipate and prevent the possibility of failure in the study. It turns out that non-academic factors also have a big influence on student success, especially family factors, such as the level of education obtained by parents, the employment status of parents and the income of both parents. To be able to understand these factors, studies are needed to predict student performance based on family background factors using machine learning models, support vector machine algorithms, naïve Bayes, neural networks and decision trees. The data used was 365 records and 11 attributes, separated by 70% for train data and 30% for test data, which was then used by kfold cross validation to evaluate the results using the parameters n_split=10 and random_state=42. In the confusion matrix parameters, the average (mean) accuracy value for the support vector machine model was 87.68%, naïve Bayes was 90.97%, neural network was 87.95% and decision tree was 85.75%. Meanwhile, the best fold result for SVM is located at the 10th fold with an accuracy of 94.44%, for NB it is located at the 4th fold with an accuracy value of 97.29%, for NN it is located at the 4th fold with an accuracy value of 94.59% and for DT is located on the 5th fold with an accuracy value of 91.89%. Thus, evaluation using k-fold cross validation can be used to predict student performance based on family attributes using the 4th fold which has the highest accuracy of 97.29% in the naïve Bayes model algorithm in order to graduate on time.]]>
Copyrights © 2024
