AI-Based Dropout Prediction and Counselling System

Abstract

Student dropout is a serious concern for higher education institutions because it affects the academic progress of students, the utilisation of institutional resources, and the overall quality of education. In many institutions, at-risk students are identified only after poor attendance, low marks, or backlogs have already accumulated, and counselling is carried out without structured data or follow-up. This work presents an AI-based dropout prediction and counselling system that combines student academic, attendance, and behavioural data with machine learning models and a PostgreSQL database. The system pre-processes the collected data, predicts the dropout probability of each student, classifies the student into a low, medium, or high risk level, and identifies the factors contributing to the prediction using an explainability technique. Based on the risk level, a counselling recommendation is generated, the concerned mentor or counsellor is alerted, and counselling sessions and follow-up actions are recorded in the same student record. The proposed framework therefore connects early identification of at-risk students with timely counselling support in a single workflow, and provides an organised basis for student retention activities in institutions.