<![CDATA[This study aims to design and implement a system based on the Decision Tree algorithm to analyze and predict the timely graduation of students at Universitas Muhammadiyah Bengkulu. The system was developed as a desktop application using Python with a Tkinter-based Graphical User Interface (GUI) and features interactive decision tree visualization. The dataset comprised 1,296 raw records of graduates from all study programs across the May 2026 and October 2025 graduation periods; following preprocessing steps—such as removing duplicate headers, standardizing date formats, and creating derived variables (GPA Category and Graduation Status)—1,144 valid records remained for analysis. The model was trained using the Gini criterion, a maximum depth of 6, a minimum sample split of 2, and "balanced" class weighting to address the data imbalance between the "Timely" and "Delayed" graduation classes, utilizing an 80:20 split for training and testing data (912 training samples and 229 test samples). Test results demonstrated an accuracy of 82.1%, precision of 89.4%, recall of 82.1%, and an F1-score of 84.3%, alongside an average 10-fold cross-validation score of 76.9% (standard deviation of 0.0622), indicating the model's stability across various data subsets. Based on the resulting decision tree structure (6 levels deep with 31 leaf nodes), the academic factors most influential on timely graduation were graduation honors and GPA category, followed by study program and gender. The application features a five-tab navigation structure—comprising Data, Model & Evaluation, Decision Tree, Confusion Matrix, and Black Box Testing tabs—and includes a function to export results to Microsoft Excel format.]]>
