Projects
Student Dropout Predictin with Decision Tree & Random Forest Algorithm
Machine Learning
This project explored the use of machine learning to predict student success and dropout. The implemented
solution utilized Decision Tree (DT) and Random Forest (RF) classification algorithms to analyze student
data. After cleaning and preparing the data, techniques were employed to handle imbalances and perform
exploratory analysis to understand the data's underlying characteristics. Both models were then built
and evaluated, with overfitting addressed in the Decision Tree. Finally, the project culminated in the
deployment of the model within a user-friendly Flask application for predicting student outcomes.
Vertical Search Engine & Subject Classifier System
Information Retrieval
This project developed a web-based, Google Scholar-like search engine focused on healthcare research by
the Research Centre for Intelligent Healthcare (RCIH). Crawling RCIH member publications weekly, the
system extracts author names, publication years, and titles, while offering keyword-based search with
relevancy ranking and subject classification functionalities to empower researchers with up-to-date and
focused exploration of RCIH's healthcare research.
Retail Intelligence System
Research & Development
Harnessing the power of data to improve retail business through data analytics. The research was
conducted using agile development methodology. The project uses various analytical & data management
techniques like concurrency management, high availability, and security to improve the efficiency of the
project. Oracle Linux 6.7 and Oracle Database 11g were setup for backend and Tableau for frontend data
visualization. Furthermore, various concepts of data warehouse were implemented to improve the
accessibility of data for analysis. Numerous ethical issues and physical disabilities like cognitive
disabilities and visual impairment were also considered during the research and development of the
project
Everest Book Store
Android & Web Application
Everest Book has both android and web applications. NodeJs-MongoDB RestAPI backs the application backend.
Vuejs Framework is used with custom CSS for the frontend web application and Kotlin for Android. The
frontend & backend were served on Vercel and Heroku. MongoDB Atlas was used for data persistency and
Cloudinary for CDN. The Web app has an Admin dashboard to edit the items and check the order list. A
small WearOS application was also implemented for the android application and various sensor triggers
were implemented on the android application.
