Analysis of Student Attendance in Information Systems Class B Cohort 2025 Using the Naïve Bayes Algorithm

Abstract

Student attendance is an important indicator for evaluating discipline and engagement in the learning process; however, manual attendance systems remain vulnerable to recording errors and data inconsistencies. This study analyzes attendance patterns of Information Systems Class B students from the 2025 cohort by applying the Gaussian Naïve Bayes algorithm. The dataset consists of 71 attendance records collected from 10 students over four lecture days, including features such as lecture day, arrival time, login activity, and prior attendance frequency. The data were processed through cleaning, encoding, normalization, and stratified train–test splitting with an 80:20 ratio.

The Naïve Bayes model achieved perfect accuracy (1.000) on the test dataset, with precision, recall, and F1-score values of 1.00 for both classes. However, this perfect performance suggests potential issues such as highly homogeneous data structure, small dataset size, or overly strong feature–label relationships, indicating possible overfitting or data leakage. This study highlights the potential of simple probabilistic classification methods for student attendance analysis while emphasizing the need for more rigorous validation using larger and more diverse datasets.