Abstract: Data mining is the process of identifying patterns, correlations, and anomalies in large data sets. Data mining involves preparing selected data sets appropriately for analysis, thus revealing expected or unexpected meaningful information. The aim of the classification method is to create a model that can predict which class or category a new data sample belongs to based on its characteristics. In our study, our dataset is a dataset obtained from the UCI Machine Learning website that examines Student Dropout and Academic Performance. This dataset focuses on various attributes such as marital status, gender, and socio-economic factors, etc. In this study, topics such as data preparation for data mining and model performance evaluation criteria are discussed. In addition, decision trees and algorithms such as XGBoost, Catboost, Naive Bayes, and AdaBoost are also examined. The data set employed here is a large collection that investigates student dropout and academic achievement. During the application phase, two separate models were compared, each trained on different splits of 80 percent and 75 percent of the data. Statistics from algorithms like Decision Trees, Random Forest, Naive Bayes, Gradient Boosting, and CatBoost were compared, and the best accuracy, 78.90 percent, was achieved by the CatBoost algorithm trained on the 75 percent split. This result led to the conclusion that the method built with CatBoost is the most effective classification approach for this problem.

Keywords: Data mining, Classification algorithms, CatBoost, Gradient Boosting, XGBoost, AdaBoost

Pages: 42-51