Machine Learning-Based Website for Student Psychological Assessment with Support Vector Machine and Rapid Application Development

Abstract

Purpose – This study aims to develop a web-based DCM assessment system that helps guidance and counseling teachers process student assessment data systematically and document follow-up needs.Methods – Rapid Application Development was used to create the system. Data were collected from 616 students using a 200-item Problem Checklist across seven domains. DCM scores and predicates were calculated using percentage-threshold rules. SVM was added as a response-pattern classification layer and compared with six other classifiers. The selected SVM model used an RBF kernel, C = 1.0, gamma = scale, one-vs-rest decision function, and probability output.Findings – Support Vector Machine achieved the best aggregate performance, with an accuracy of 0.8871, precision of 0.884993, recall of 0.887097, and F1-score of 0.883355. Usability testing produced a System Usability Scale score of 86.72, while User Acceptance Testing with three guidance and counseling teachers reached a 100% success rate.Research implications – The system combines deterministic DCM scoring and data-driven response-pattern classification to support counselor review of student assessment records and documented follow-up considerations. Because the evaluation used one school, imbalanced labels, and no actual predicate A samples, the model output should be used as a decision-support indicator and not as evidence of improved counseling outcomes, workload reduction, service acceleration, or prioritization accuracy.Originality – This study integrates rule-based DCM scoring, SVM-based response-pattern classification, and a web-based school counseling workflow in the Indonesian secondary school context.