Prediction of undergraduate major selection via machine learning algorithms

In modern world, making a decision on a major for high school students can be a challenging step in their lives. Especially, if students are not aware of their potential talents, interests, personalities and goals to reach in the future. Therefore, students search and look for various sources in order to obtain the best objective and scientific perspective on their major preferences. However, advices from some guidance services or recommendations from friends and close relatives who graduated from college may not be objective and helpful. This is because these recommendations are shaped with guiders' preferences than taking into consideration of the student's ability. In this context, this study aims to develop an application that uses machine learning algorithms to assist students in their university major choices. The application allows students to predict which departments they may be successful by using their achievement scores in high school Mathematics, Physics, Chemistry, Biology, Geography, Turkish, History, and foreign language courses. The data obtained in our study were collected from all regions of Turkey, cleaned, and transformed into a dataset to ensure more consistent and generalizable learning models. In the study, different learning models are used such as XGBoost, SVC, KNN, and Gaussian Naive Bayes learning models. Among these models, the highest performance was achieved with the KNN algorithm. This application aims to be an important step in reducing the difficulties students face in their university major choices. Because this application will provide students with an objective and scientific perspective, helping them predict which departments they may be successful without being influenced by personal preferences or environmental factors. It is aimed that this study can inspire future research and be a useful resource for students struggling with university major choices.