Final Year Project

Student Dropout Risk Prediction

A machine learning-based predictive analytics project for identifying student withdrawal risk in higher education using academic and demographic indicators.

Live Streamlit Demo

Predictive analytics for student retention.

This project uses a cleaned higher education student dataset, supervised machine learning, class imbalance handling, and a deployed Streamlit dashboard to support early visibility into dropout-related risk patterns.

Project Type Machine Learning Dashboard
Current Status Deployed Demo
Core Stack Python · Scikit-learn · Streamlit

Project Summary

This final year project focuses on using machine learning to predict student dropout risk in higher education. The work includes dataset preparation, target encoding, feature processing, class imbalance handling, model training, evaluation, and dashboard deployment through Streamlit Community Cloud.

Problem Statement

Higher education institutions need earlier visibility into students who may be likely to withdraw, so intervention and support can be planned before the issue becomes final. This project explores how predictive analytics can support that decision-making process.

Dashboard Purpose

The deployed dashboard makes the model easier to understand by allowing users to explore student-level inputs, prediction outputs, and supporting visual insights in a simple web interface.

Technical Workflow

01 Data Preparation

Cleaned the dataset, handled missing values, fixed data types, and prepared features.

02 Model Training

Built classification models for identifying withdrawal risk from selected student attributes.

03 Evaluation

Reviewed performance using classification metrics suitable for imbalanced prediction.

04 Deployment

Deployed a Streamlit dashboard so the model can be demonstrated interactively.

Skills Demonstrated

Python Data Cleaning Machine Learning Classification SMOTE / Imbalance Handling Streamlit Deployment