CORS (Course Offering Recommendation System) is an AI-powered academic planning platform built for university administrators and academic planners at the Lebanese American University (LAU). It analyzes your institution’s historical enrollment data and applies trained machine learning models to recommend which courses to offer each semester, how many sections to open, and how to schedule them — all from a single dashboard.Documentation Index
Fetch the complete documentation index at: https://cors-lau.vercel.app/docs/llms.txt
Use this file to discover all available pages before exploring further.
Who CORS is for
CORS is designed for academic planners, department chairs, and university administrators who are responsible for building each semester’s course schedule. If you make decisions about which courses to run, how many sections to open, and how to allocate faculty and classrooms, CORS is built for you.Academic planners
Generate AI-backed recommendations for course offerings and section counts based on historical demand.
Department chairs
Identify prerequisite bottlenecks and high-demand courses before the scheduling window opens.
Scheduling coordinators
Assign time slots, rooms, and faculty in the built-in timetable scheduler and export the final schedule.
Data administrators
Manage course catalogs, historical enrollment records, and faculty rosters that power the ML models.
Key capabilities
CORS brings four core capabilities together in one platform: AI course recommendations — Machine learning models score each course in your catalog and recommend whether it should be offered next semester, along with a projected section count. Courses are labeled either Recommended or De-prioritized based on historical enrollment patterns, failure rates, and projected student demand. Prerequisite graph — An interactive visualization of course dependencies across your curriculum. You can explore which courses are prerequisite bottlenecks — courses that block large numbers of students from progressing — and factor that into your offering decisions. Timetable scheduler — A drag-and-drop scheduler for building the semester timetable once you have your course list. CORS detects time conflicts for rooms and faculty automatically, so you can finalize a conflict-free schedule before exporting. Data management — A centralized area for uploading and managing the three datasets that power every analysis: your course catalog, historical offering records, and faculty roster.Supported campuses
CORS supports both LAU campuses with independent ML models trained on each campus’s own enrollment history:Beirut campus
A dedicated ensemble model trained exclusively on Beirut enrollment history, calibrated to Beirut-specific demand patterns.
Byblos campus
A dedicated ensemble model trained exclusively on Byblos enrollment history, calibrated to Byblos-specific demand patterns.
How CORS works
At a high level, CORS follows a five-stage workflow from raw data to an exported schedule:Upload historical data
Import your course catalog (JSON), historical enrollment and offering records (CSV or XLSX), and faculty records (JSON, CSV, or XLSX) through the Data Management page. This data is the foundation of every analysis.
ML models score courses
When you trigger an analysis, CORS runs your data through the campus-specific machine learning model. The model evaluates each course using metrics such as historical failure rates and enrollment trends, then assigns a confidence score to each course.
Recommendations generated
CORS surfaces a ranked results table. Each course is labeled Recommended (offer it) or De-prioritized (skip it), along with a suggested section count and supporting metrics.
Build the schedule
Take your recommended courses to the Timetable Scheduler and assign each section a time slot, room, and faculty member. CORS flags any conflicts in real time.
Dashboard metrics
After your first analysis run, the CORS dashboard shows four key metrics that give you a snapshot of your data and model health:| Metric | What it shows |
|---|---|
| Course catalog | Total number of validated courses in the current academic year registry |
| Historical data | Total processed enrollment records across all departments |
| Model ROC-AUC | The area under the ROC curve — a measure of how well the model distinguishes between courses that should and should not be offered |
| Average failure rate | Historical percentage of failing grades across all recorded course offerings |
The Model ROC-AUC score displays as N/A until you run your first analysis. Once an analysis completes, the score reflects the global average confidence across the most recent run.
Next steps
Quickstart
Follow a step-by-step walkthrough to run your first analysis.
Authentication
Learn how to log in and authenticate API requests.