Analysis of Informatics Engineering Students’ Dependency Level on the Use of ChatGPT Using the Support Vector Machine Method

Aswita Indah Luthfiana Hasibuan (Universitas Malikussaleh)
Dahlan Abdullah (Universitas Malikussaleh)
Kurniawati Kurniawati (Universitas Malikussaleh.)

International Journal of Engineering, Science and Information Technology
Vol 6, No 2 (2026)

This study aims to analyze the level of dependency among Informatics Engineering students on the use of ChatGPT in academic activities using the Support Vector Machine (SVM) method. The research data were collected through questionnaires distributed to students of the Informatics Engineering Study Program at Universitas Malikussaleh from the 2022–2025 cohorts, involving a total of 400 respondents. The research indicators included usage intensity, duration of use, purpose of use, perceived effectiveness, and dependency behavior toward ChatGPT. The collected data underwent several preprocessing stages. including missing value checking. label transformation. data normalization using Min-Max Scaling. and dataset splitting into 80% training data and 20% testing data. The classification process was performed using the Support Vector Machine (SVM) algorithm with a linear kernel. The experimental results showed that the proposed SVM model successfully classified student dependency levels into three categories, namely low, medium, and high, achieving an accuracy of 95%, which indicates excellent classification performance. The findings also revealed that usage frequency, duration of use, and the utilization of ChatGPT for academic assignments and programming activities were the most influential factors affecting student dependency. Furthermore, a web-based system was developed using Python Flask and SQLite to facilitate data processing, model training, and visualization of classification results. The system testing results demonstrated that all implemented features functioned properly according to the specified requirements.