Ellery (Backend)

Your student university portal, made richer and smarter

Ellery is a full-stack web platform that allows university students to explore their student portal with a richer and more intuitive platform. In a way, Ellery becomes your personal college path assistant. It is composed of a decoupled backend JSON API and a front-end web application.

In our case, we created a demonstration of Ellery interfacing with our university's MyFIU student portal, but our platform can interface with any university platform in practice. It works by scraping a student's dataset from their portal upon giving us their student portal credentials.

Statistical Analysis

Our backend utilizes data gathered from transcripts and online grade distributions present on your student portal. Using this, we determine how well you will do in specific courses and on your major as a whole. Currently, our statistical inference only works for Computer Science majors (because of time constraints, but the applications are endless).

Under the hood, our backend passes a JSON file of the student's academic data to the analysis engine that does our predictions. A sample of the passed data looks like this:

{
    "semesters": [{
        "semester": "fall",
        "year": "2012",
        "termGpa": "4.0",
        "cumulativeGpa": "3.5",
        "courses": [{
            "id": "CHM 1045", 
            "credits": "3.0",
            "creditsEarned": "9.0",
        }, {
            "id": "COP2210",
            "credits": "3.0",
            "creditsEarned": "12.0"
        }]
    }]
}

The analysis engine returns this, which the backend API shuttles forward to the front-end:

{
    "expectedGpa": "3.0", 
    "recommendedCourses": ["CHM 1045"]
}

API Endpoints

All API endpoints have the prefix api/<elleryVersion>. All the API endpoints return an output that has the form:

{
    "status": "<int: httpStatusCode>",
    "message" "<string: additional request/response information">,
    "data": "<?: whatever the API endpoint should return>"
}

1. /login
   • Given FIU credentials, return a JWT Token with the user's type (permissions), university email address, system supplied unique identifier and name.
     • POST
     • Accepts {username, password}
     • Returns {jwtToken}
2. /student
   • All these API endpoints require either an:
     • Authorization: Bearer <jwtToken> provided as the Authorization header OR
     • token=<jwtToken> as part of the request parameters
   • GET /courses/present
     • Returns a list of all courses that a student is currently taking
     • Returns [{courseType, courseId, sectionNumber, instructor, schedule:[{days, time}], location}]
   • GET /courses/all
     • Returns a list of all courses that a student has taken up to now
     • Returns [{courseType, courseId, sectionNumber, instructor, schedule:[{days, time}], location}]
   • GET /prediction
     • Returns a personalized prediction set of your academic path, using our analysis engine
     • Returns {expectedGpa, recommendedCourses:[]}

Docker Environment

A Docker image of our entire backend's environment is available at https://hub.docker.com/r/alastairparagas/ellery_web. This includes all of the environment dependencies needed to run ellery, from the R statistics package, to Node, PhantomJS and CasperJS.

By default, the Docker container is exposed at port 8000. Mount the codebase volume at /source. Make sure that when running the container, to run the backend app's entrypoint as node /source/<entryFileHere>.