A university wants to identify students who are at risk of failing or dropping out before the end of the semester. The objective of this competition is to develop a machine learning model that predicts whether a student will successfully complete a course.
Participants must analyze student behavioral and academic data collected through Learning Management Systems (LMS), attendance records, assessments, and student engagement activities. The goal is to build predictive models that identify at-risk students early enough for intervention.
Objectives Participants will demonstrate their ability to: - Clean and preprocess data - Perform exploratory data analysis (EDA) - Engineer predictive features - Train machine learning models - Evaluate classification models - Interpret results using Explainable AI - Communicate findings
Dataset Description The training dataset contains 10,000 student records with a target variable included. Key variables include Student_ID, Age, Gender, Major, GPA_Previous, Attendance_Rate, LMS_Login_Count, Discussion_Posts, Assignment_Submission_Rate, Avg_Assignment_Score, Midterm_Score, Hours_Studied_Per_Week, Financial_Aid, Employment_Status, and Final_Status (Pass=1, Fail=0). The test dataset consists of 3,000 student records where participants must predict the Final_Status.
Evaluation
The primary metric is the F1 Score, chosen because it handles class imbalance and is suitable for educational intervention systems. Participants must submit a submission.csv file containing Student_ID and predicted Final_Status.
Deliverables In addition to the prediction file, participants must submit a technical report (maximum 10 pages) covering an Executive Summary, Data Cleaning, EDA, Feature Engineering, Model Selection, Results, Interpretation, and Recommendations. A Jupyter Notebook including data preprocessing, model training, evaluation, and visualizations is also required.
Prizes Total Prizes Available: 3 - 1st Place: $300 - 2nd Place: $200 - 3rd Place: $100