Course Recommendation System

Intelligent Content-Based Course Discovery Engine Powered by NLP & Machine Learning

This Python Flask web application delivers personalized course recommendations across a curated catalog of 3,424+ online courses. It leverages Natural Language Processing (TF-IDF Vectorization and Cosine Similarity) coupled with diversity-aware candidate selection to guide students and professionals toward tailored educational pathways.

Key Features

Personalized Course Recommendations: Generates tailored course suggestions based on textual similarity across thousands of course descriptions using TF-IDF vectorization and cosine similarity.
Diversity-Aware Candidate Selection: Balances similarity scores with institutional and difficulty diversity to avoid redundant recommendations from the same university or skill level.
Interactive Catalog Search: Enables users to search a catalog of 3,400+ online courses by keyword or subject with real-time autocomplete suggestions and live course counters.
Quick-Topic Exploration: Provides one-click discovery for popular domains including Python, Data Science, Machine Learning, Web Development, Artificial Intelligence, and Cloud Computing.
Comprehensive Course Profiles: Displays essential course information including offering university/partner, difficulty tier, learner rating, technical skills taught, and expandable course syllabus descriptions.
Direct Course Enrollment Access: Features verified direct links to original Coursera course pages for streamlined access and enrollment.
Customizable Recommendation Count: Allows users to specify how many recommendations to generate, toggling between Top 3, 5, 8, or 10 courses.
RESTful API Backend: Exposes dedicated endpoints (/courses, /recommend, /api/health) for programmatic access, enabling seamless integration with external clients or frontends.

Dataset & Data Source

The recommendation engine is built on the Coursera Course Dataset, publicly available on Kaggle:

Source Platform: Kaggle - Coursera Course Dataset
Author / Publisher: Siddharth M
Dataset Size: 3,424 cleaned and structured course records
Format: CSV stored at data/cleaned_dataset.csv

Included Features:

Column Name	Description	Example
`Course Name`	Full title of the course on Coursera	Python For Everybody
`University`	Partner university or organization offering the course	University of Michigan
`Difficulty Level`	Course target level (`Beginner`, `Intermediate`, `Advanced`, `Conversant`)	Beginner
`Course Rating`	Average user rating on Coursera (scale 1.0 - 5.0)	`4.8`
`Course URL`	Direct URL to access course materials on Coursera	`https://www.coursera.org/learn/...`
`Course Description`	Text description covering syllabus, topics, and objectives	Learn fundamental syntax...
`Skills`	Technical domain competencies and tags	Python, JSON, Data Structures

Recommendation Engine Pipeline

flowchart TD
    A["User Inputs Keyword / Selects Topic"] --> B["API: GET /courses?keyword=..."]
    B --> C["Filter Catalog by Course Title"]
    C --> D["User Selects Reference Course"]
    D --> E["API: GET /recommend?course_title=...&n=5"]
    E --> F["Retrieve Precomputed TF-IDF Cosine Similarity Vector"]
    F --> G["Candidate Ranking & Diversity Pool (Universities & Difficulties)"]
    G --> H["Enrich with Clean Metadata (Institution, Rating, Skills)"]
    H --> I["Return JSON Response & Render Responsive UI Cards"]

How the Model Works

TF-IDF Vectorization: Course descriptions are processed with n-gram ranges (1, 2) and English stop words filtered to capture technical concepts and domain terms.
Cosine Similarity: Measures the angular distance between course vectors, capturing nuanced conceptual relevance.
Diversity-Aware Selection: Balances high similarity with variety across different institutions and difficulty tiers to avoid repetitive recommendations.

Project Structure

Course-Recommendation-System/
├── app.py                          # Flask application, routing & recommendation service
├── model.py                        # Training & serialization script
├── requirements.txt                # Production Python dependencies
├── requirements-dev.txt            # Development & testing dependencies (pytest, pytest-cov)
├── pytest.ini                      # Pytest runner configuration
├── package.json                    # Node.js configuration & test scripts
├── playwright.config.js            # Playwright E2E configuration & webServer setup
├── data/
│   └── cleaned_dataset.csv         # Raw catalog dataset (3,424 Coursera courses)
├── models/
│   ├── cosine_sim.pkl              # Precomputed pairwise cosine similarity matrix
│   ├── df.pkl                      # Serialized course DataFrame
│   ├── tfidf_vectorizer.pkl        # Fitted TF-IDF vectorizer
│   ├── svd_model.pkl               # Collaborative filtering SVD model
│   └── knn_model.pkl               # Collaborative filtering KNN model
├── static/
│   ├── css/styles.css              # Modern stylesheet with variables, badges & animations
│   └── js/scripts.js               # Client-side debounce, API calls, and card rendering
├── templates/
│   └── index.html                  # Modern Bootstrap 5.3 HTML5 template
└── tests/
    ├── unit/
    │   └── test_recommendation_logic.py # Algorithmic & logic unit tests (12 tests)
    ├── integration/
    │   └── test_api_routes.py           # Flask API client integration tests (11 tests)
    └── e2e/
        └── course-recommendation.spec.js # Playwright E2E browser test suite (8 scenarios)

Test Suite Matrix

Suite	Category	Runner & Language	File	Tests	Coverage Scope
Logic Unit Tests	Backend Logic	`pytest` (Python)	`tests/unit/test_recommendation_logic.py`	12	Ranking accuracy, self-exclusion, schema validity, title casing, boundary handling, mock pipeline, keyword matching
Integration Tests	API Routes	`pytest` (Python)	`tests/integration/test_api_routes.py`	11	Status codes (200, 400, 404), schema verification, param parsing (`n`), query validation, `/api/health`
Playwright E2E	Full Browser	Playwright (JavaScript)	`tests/e2e/course-recommendation.spec.js`	8	Initial UI render, debounced autocomplete, topic pills, card generation, count change, form clear, empty validation, mobile layout
TOTAL				31	Full end-to-end coverage

Setup & Installation

1. Prerequisites

Python 3.10+ (Tested on Python 3.12)
Node.js 18+ & npm

2. Clone the Repository

git clone https://github.com/saboye/Course-Recommendation-System.git
cd Course-Recommendation-System

3. Set Up Python Virtual Environment

# Create virtual environment
python -m venv venv

# Activate on Windows:
venv\Scripts\activate

# Activate on macOS / Linux:
source venv/bin/activate

# Install application and testing dependencies:
pip install -r requirements.txt
pip install -r requirements-dev.txt

4. Install Node.js & Playwright Dependencies

npm install
npx playwright install chromium

Running the Application

Start the Flask server:
```
python app.py
```
Access the application: Open your browser and navigate to:
```
http://127.0.0.1:5000/
```

Executing Tests

Run All Python Tests (Unit & Integration)

pytest

Run Logic Unit Tests Only

pytest tests/unit

Run API Integration Tests Only

pytest tests/integration

Run Python Tests with Code Coverage Report

pytest --cov=app tests/

Run Playwright End-to-End Tests (JavaScript)

# Run headless browser tests
npm test

# Run in headed (visible) browser mode:
npm run test:e2e:headed

# Open interactive Playwright UI Runner:
npm run test:e2e:ui

API Reference

1. `GET /api/health`

Health check and system status.

Example Request:
```
curl http://127.0.0.1:5000/api/health
```

Example Response (200 OK):

{
  "has_models": true,
  "status": "healthy",
  "total_courses": 3424
}

2. `GET /courses`

Search the course catalog by keyword.

Parameters: keyword (string, required)

Example Request:

curl "http://127.0.0.1:5000/courses?keyword=Python"

Example Response (200 OK):

{
  "count": 48,
  "courses": [
    "Python For Everybody",
    "Applied Data Science With Python",
    "Crash Course On Python"
  ]
}

3. `GET /recommend`

Get content-based course recommendations based on a reference course.

Parameters:
- course_title (string, required): Title of reference course.
- n (integer, optional, default: 5, range: 1-20): Number of recommendations to return.

Example Request:

curl "http://127.0.0.1:5000/recommend?course_title=Python%20For%20Everybody&n=3"

Example Response (200 OK):

{
  "count": 3,
  "query": "Python For Everybody",
  "selected_course": {
    "Course Name": "Python For Everybody",
    "University": "University Of Michigan",
    "Difficulty Level": "Beginner",
    "Course Rating": 4.8
  },
  "recommendations": [
    {
      "Course Name": "Python Data Structures",
      "University": "University Of Michigan",
      "Difficulty Level": "Beginner",
      "Course Rating": 4.9,
      "Course URL": "https://www.coursera.org/learn/python-data",
      "Course Description": "This course will introduce the core data structures of the Python programming language...",
      "Skills": "python programming tuples python dictionaries json"
    }
  ]
}

Troubleshooting & FAQ

1. PowerShell script execution disabled error when activating venv

If you receive an execution policy error in Windows PowerShell, run:

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope Process
venv\Scripts\activate

2. Playwright cannot find browser executable

Install the browser binaries using:

npx playwright install chromium

3. Port 5000 is already in use

If port 5000 is occupied, you can stop the conflicting process or change the port in app.py and playwright.config.js:

app.run(host='127.0.0.1', port=5001, debug=True)

License

This project is licensed under the MIT License - see the LICENSE file for details.

Navigation Menu

Repository files navigation

Course Recommendation System

Table of Contents

Key Features

Dataset & Data Source

Recommendation Engine Pipeline

How the Model Works

Project Structure

Test Suite Matrix

Setup & Installation

1. Prerequisites

2. Clone the Repository

3. Set Up Python Virtual Environment

4. Install Node.js & Playwright Dependencies

Running the Application

Executing Tests

Run All Python Tests (Unit & Integration)

Run Logic Unit Tests Only

Run API Integration Tests Only

Run Python Tests with Code Coverage Report

Run Playwright End-to-End Tests (JavaScript)

API Reference

1. `GET /api/health`

2. `GET /courses`

3. `GET /recommend`

Troubleshooting & FAQ

License

About

Releases

Packages

Used by

Contributors

Languages

Name		Name	Last commit message	Last commit date
.vscode		.vscode
data		data
models		models
notebooks		notebooks
static		static
templates		templates
tests		tests
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
README.pdf		README.pdf
app.py		app.py
model.py		model.py
package-lock.json		package-lock.json
package.json		package.json
playwright.config.js		playwright.config.js
pytest.ini		pytest.ini
requirements-dev.txt		requirements-dev.txt
requirements.txt		requirements.txt

Latest commit

History

Folders and files

Repository files navigation

Course Recommendation System

Table of Contents

Key Features

Dataset & Data Source

Recommendation Engine Pipeline

How the Model Works

Project Structure

Test Suite Matrix

Setup & Installation

1. Prerequisites

2. Clone the Repository

3. Set Up Python Virtual Environment

4. Install Node.js & Playwright Dependencies

Running the Application

Executing Tests

Run All Python Tests (Unit & Integration)

Run Logic Unit Tests Only

Run API Integration Tests Only

Run Python Tests with Code Coverage Report

Run Playwright End-to-End Tests (JavaScript)

API Reference

1. GET /api/health

2. GET /courses

3. GET /recommend

Troubleshooting & FAQ

License

About

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Used by

Contributors

Languages

1. `GET /api/health`

2. `GET /courses`

3. `GET /recommend`