Course Recommendation System

This project is a content-based course recommendation system that suggests Udemy courses based on user input keywords. The system leverages TF-IDF vectorization and cosine similarity to provide relevant course recommendations.

Table of Contents

• Dataset
• Installation
• Project Structure
• Usage
• Model and Features
• Running the Streamlit App
• License

Dataset

The dataset used for this project is sourced from Kaggle: Udemy Course Dataset.

Important: Please check the licensing information on the Kaggle page before using this dataset in production.

Installation

To set up the project, follow these steps:

# Clone the repository
git clone https://github.com/yourusername/course-recommendation-system.git
cd course-recommendation-system

# Create a virtual environment using Conda
conda create --name course_recommendor python=3.12 -y
conda activate course_recommendor

# Install dependencies
conda install --file requirements.txt

Project Structure

├── artifacts/
│   ├── udemy_courses.csv             # Raw dataset
│   ├── cleaned_udemy_courses.csv     # Processed dataset
│   ├── tfidf_vectorizer.pkl          # Saved TF-IDF model
│   ├── similarity_matrix.pkl         # Saved similarity matrix
├── src/
│   ├── data_preprocessing.py         # Cleans and preprocesses data
│   ├── feature_extraction.py         # TF-IDF vectorization
│   ├── recommendation.py             # Course recommendation logic
│   ├── app.py                        # Streamlit web app
├── requirements.txt                   # Required Python libraries
├── environment.yml                    # File to set up environment directly using YAMl file
├── README.md                          # Project documentation

Usage

Step 1: Data Preprocessing

Run the following command to clean the dataset:

python src/data_preprocessing.py

This script:

• Combines course_title and subject columns.
• Removes stopwords and special characters.
• Saves the cleaned dataset as cleaned_udemy_courses.csv.

Step 2: Feature Extraction

Extract features using TF-IDF:

python src/feature_extraction.py

This script:

• Converts text data into TF-IDF vectors.
• Saves the tfidf_vectorizer.pkl and similarity_matrix.pkl files.

Step 3: Generate Recommendations

Run the recommendation script to test:

python src/recommendation.py "machine learning"

This will return a list of recommended courses based on the keyword provided.

Running the Streamlit App

To start the web application:

streamlit run src/app.py

This will open a user-friendly interface where users can enter keywords and receive recommended courses.

License

This project is for educational purposes only. Please review the Kaggle dataset license before using it in production.

---

Feel free to modify the project according to your needs! 🚀