Application of Binary Logistic Regression Method Using Python to Analyze the Effect of Ease of Use, Administrative Costs and Preferences on the Probability of Student Decisions Choosing Payment Channels for Course Registration Using Bank Mandiri

Suryani, Rahmah
Praditya, Rizqy Gumilar
Sembodo, Giry
Heikal, Jerry

Innovative: Journal Of Social Science Research
Vol. 4 No. 3 (2024): Innovative: Journal Of Social Science Research (Special Issue)

This study analyzes the factors that influence Open University students' decision to choose Bank Mandiri payment channel for course registration. Using binary logistic regression method with Python, this study explores the influence of ease of use, administrative costs, and student preference on the probability of choosing Bank Mandiri. Data from 200 students were analyzed using variables such as age, gender, ease of use of mobile banking application, transaction fees, and student preference. Model evaluation using confusion matrix and ROC curve showed good performance with Area Under the Curve (AUC) 0.84. This study revealed that factors such as ease of use and transaction fees strongly influence students' decisions in choosing payment channels. The results show that the model has 78% accuracy with three significant variables with values below 5%: ease of use, transaction costs, and student preferences. Evaluation using ROC curves showed good model performance with an AUC of 0.84. This research aims to improve UT and Bank Mandiri payment services and provide insight into the adoption of financial technology in the education sector.