An intelligent and interactive recommendation engine built using R and MongoDB, designed to suggest personalized product recommendations in an e-commerce setting.
This project was developed as part of the Big Data Analytics course at College of Engineering, Guindy (Anna University). The system delivers dynamic, personalized product recommendations using multiple filtering techniques and presents insights through an interactive Shiny Dashboard.
Our system provides personalized product recommendations using:
-
🧠 Collaborative Filtering
Uses user-item interactions to recommend products based on similarities between users (User-based) or between items (Item-based). -
📄 Content-Based Filtering
Leverages product metadata like descriptions, tags, and names to recommend products similar to what a user has liked or searched for. -
🎯 Constraint-Based Filtering
Allows filtering recommendations by price range and minimum rating to align with user preferences. -
🔗 Hybrid Approach
Combines collaborative and content-based filtering scores to enhance both diversity and accuracy of recommendations.
Recommendation systems are crucial in enhancing user experience across digital platforms — from suggesting the next movie on Netflix to recommending products on Amazon. In e-commerce, such systems:
- Improve customer engagement
- Increase conversion rates
- Drive sales and retention
With the global e-commerce recommendation engine market expanding rapidly, building smart, scalable recommendation systems is more relevant than ever.
We used the Walmart E-Commerce Product Dataset from Kaggle, containing:
- Product metadata: ID, name, brand, price, rating, description, image URLs
- User profiles: ID, age, gender, location
- Interaction data: user-product ratings and timestamps
This provided a rich, real-world foundation for experimenting with and optimizing our recommendation algorithms.
-
💻 R Programming Language
shiny,shinydashboard: For interactive UIrecommenderlab,proxy,tm: For recommendation logic and text processingmongolite,dplyr,stringr,reshape2: For database and data manipulation
-
🗃️ MongoDB (NoSQL)
Used to store users, product data, and interaction history. -
🧪 Shiny Dashboard
A responsive web interface to explore recommendations, filter options, and visualize insights.
In addition to recommendation features, the dashboard includes data visualizations to help understand:
- Distribution of ratings
- Most popular products
- User activity patterns
- Price vs rating trends
(i) Price Filter – Recommends products in ascending order of price
(ii) Rating Filter – Recommends products in descending order of ratings
Interacted data stored in MongoDB
- Surendiran M
- Ram Prasath K V
- Lokesh Kannan M
This project is for academic and learning purposes.
Feel free to explore, fork, and build on this system!