Educational institutions collect large amounts of attendance data every semester, yet this information is rarely used to forecast future classroom attendance. Attendance patterns are influenced by multiple academic and scheduling factors such as lecture timing, day of the week, subject, examinations, holidays, faculty schedules, and previous attendance trends. Accurate prediction of attendance can help institutions optimize classroom utilization, improve academic planning, and identify lectures that may require proactive intervention.
In this competition, participants are challenged to develop machine learning models that predict classroom attendance using historical attendance records and academic scheduling information. Competitors are encouraged to perform exploratory data analysis, engineer meaningful features, and compare different machine learning algorithms to build accurate predictive models. The competition welcomes both traditional machine learning and deep learning approaches.
This competition provides an excellent opportunity to work with real-world educational data, apply predictive analytics, and gain hands-on experience in feature engineering, regression, classification, model evaluation, and reproducible machine learning workflows.
Your task is to predict classroom attendance for future lectures using historical attendance data and academic schedule information. Depending on the competition configuration, the prediction target may be: - Attendance Percentage (Regression) - Number of Students Present (Regression) - Attendance Category (Classification)
Participants are encouraged to explore innovative feature engineering techniques and build models that generalize well to unseen academic schedules.
Reliable attendance prediction can help educational institutions: - Improve academic planning and scheduling - Identify lectures likely to experience low attendance - Optimize classroom and faculty resource allocation - Support data-driven decision making - Analyze factors affecting student attendance - Build intelligent educational analytics systems
Whether you are new to machine learning or an experienced data scientist, this competition offers a practical opportunity to apply predictive modeling techniques to solve a real-world educational problem.
Submissions are evaluated using Mean Absolute Error (MAE). A lower MAE indicates better predictive performance. For each lecture in the test dataset, participants must predict the Attendance Percentage. The predicted values are compared with the actual attendance percentages using Mean Absolute Error.
The dataset contains historical classroom attendance records along with academic scheduling information such as lecture timing, day of the week, subject, semester, branch, faculty, classroom details, and other contextual variables that may influence attendance.