The Comparison of C4.5 and CART (Classification and Regression Tree) Algorithm in Classification of Occupation for Fresh Graduate

Abstract

. The problem that college students face is the difficulty of determining the appropriate field of work after they graduate from college. In this study, a classification of the field of work was carried out using the data mining method based on the alumni field of work data. The data on the field of work of alumni contained information such as gender, study program, practical work topics, types of practical work companies, final project topics, and year of graduation. The classification on the field of work carried out was divided into three types of experiments, namely experiments in eight target categories (STQA Engineer, Software and Mobile Application Developer, Web Developer, UI/UX Designer, Software and Business Analyst, Lecturer and Researcher, AI Engineer, DevOps and Cybersecurity Practitioner), three target categories (SQA, Programmer, Data Manager, and Analyst) and two target categories (Programmer and Non-Programmer). The data mining algorithms used to classify were C4.5 and CART (Classification and Regression Tree). The accuracy obtained using the C4.5 algorithm was 42% in the eight categories experiment, 58% in the three categories experiment, and 75% in the two categories experiment. In comparison, the accuracy obtained using the CART algorithm was 43% in the eight categories experiment, 61% in the three categories experiment, and 77% in the two categories experiment. Based on the experimental results, it can be concluded that the best algorithm to classify the fields of work based on alumni data from the two algorithms used is the CART algorithm, even though the difference is not too significant.