kjartank13 / centrisretention Goto Github PK

Info about Centris Retention:

This is the code which I (and my 3 project partners) wrote for our Bachelors project in Computer Science at the Reykjavík University.

Our project was a contribution to an overall larger project called Centris. Centris is an in-development inner web system for students and faculty. The system has two parts to it. The LMS (Lecture Management System) part and the SMS (Student Management System) part. The LMS allows students to find anything pertaining to their studies, such as their timetable, assignments, lectures, grades, etc. It also allows teachers to do things like posting assignments, grades, lectures, etc. The SMS is for the administrative staff, allowing them to manage registration, scheduling, etc.

This Bachelors project, Centris Retention, is a system which is intended to follow students' activity on the LMS site, and based on that, it evaluates the probability of each individual student dropping out of their program. A list of students, with their dropout probabilities is displayed on a page in the SMS part of Centris.

Our project essentially had two parts to it: The backend and the frontend. At the start of a semester, the backend fetches a list of students and their class registrations. The backend is designed to run on a server 24/7 and every night when traffic to the LMS site is thought to be low, it will collect a log of all activities for the past 24 hours from a message queue (RabbitMQ) and start to sort through them all, calculating new dropout probabilites for every student. The frontend is a part of the SMS site. It makes API calls to the backend to get a list of all students, and their details. When a student is clicked, it is possible to graphically view the progression of their dropout risk factor for the previous month, along with brief explanation for changes in risk. It is also possible for a user (i.e. employee of the administrative office) to log when and if they contact a student offering help or warning them of their dropout risk.

This is a pretty extensive project, the backend was entirely built from scratch while the frontend was developed as part of the larger SMS site code base. Therefore this was a fine experience of both starting a large project from square one, while also getting a chance to contribute to an overall larger, already existing project. It is worth noting that this was intented to be a project which will be handed over to a new group down the line. Our goal was simply to set up the framework for this system. Since student dropout is such a nebulous and personality-based concept, predicting it in any real sense would have to be done via artificial learning intelligence. That in itself is complex enough to be it's own Bachelors project, so that is why that wasn't something that we could get into implementing.

It is also worth noting that the actual repository for this project is located elsewhere, in the greater context of the overall Centris project. However, since there's a lot of code there which I don't have the rights to, I can not share that openly. I am simply uploading the code which my group wrote and delivered upon completion of the project.