Comparison of Support Vector Regression and Artificial Neural Network Algorithm in Predicting the Number of Prospective Student Registrans

The candidate of students have difficulty in determining which college to choose. This difficulty is due to the lack of information about the college to be addressed. One of the actions that can be taken to introduce the campus is bypromoting to high schools. To find out schools that need an in-depth introduction to the X campus, it is necessary to make a prediction system that can show the possibility of the number of registrants from these schools that are still small. In this study, a comparison between the SVR and ANN methods was carried out to determine the best model in predicting the number of prospective applicants from each school. In the process of model formation, the optimal hyperparameter is found in the model using gridsearch. To determine the best model, researchers used the smallest RMSE value obtained from both models. Based on the RMSE results obtained, the ANN model RMSE value is smaller with a value of 0.04929599755078 so this model is the best model. However, after calculating MAAPE, the ANN model has not been able to predict well.