Overview
A personal AI and machine-learning project that helps students decide what to do next. It starts from academic data and moves step by step from data processing to analysis, a prediction model, a recommendation engine and finally a web application. This entry describes the roadmap, not finished work.
Intended outcome
A student logs in, enters academic results and receives analysis, predictions and revision recommendations, each with an explanation of why it appears.
Planned phases
- Phase 1, foundation: Python file handling, Git and GitHub, and a program that reads and validates a small academic dataset.
- Phase 2, data analysis: NumPy, Pandas and Matplotlib to find patterns in performance and explain what the data does and does not show.
- Phase 3, first model: scikit-learn regression with a baseline, a training/testing split and honest limits.
- Phase 4, recommendation engine: transparent rules that link revision advice to specific topics.
- Phase 5, web application: Django and PostgreSQL with accounts, student profiles, a dashboard and a recommendations page.
- Phase 6, test, secure and deploy: automated tests, input validation, protected data and documentation.
- Phase 7, evaluation: a small group of willing students, with informed consent, to test it.
- Phase 8, scale: only features that solve a demonstrated problem.
Planned technologies
Python, NumPy, Pandas, Matplotlib and scikit-learn for the data and model work; HTML, CSS and JavaScript with Django and PostgreSQL for the web application; Git and GitHub for version control.
How I plan to work
- Finish one phase before starting the next, and test every major feature before building on it.
- Understand each library before depending on it, and use AI assistants to explain, review and debug code rather than to generate the whole application.
- Protect students’ data and get permission before collecting real records.
- Judge progress by working deliverables and demonstrated understanding, not by the amount of code written.
Status
Planned. No phase is presented here as complete.