https://github.com/philiptitus/book-recommendation
Recommender system using collaborative filtering algorithm for recommnding books to read
https://github.com/philiptitus/book-recommendation
adam-optimizer bookrecommendsystem collaborative-filtering recommender-system unsupervised-machine-learning
Last synced: 2 months ago
JSON representation
Recommender system using collaborative filtering algorithm for recommnding books to read
- Host: GitHub
- URL: https://github.com/philiptitus/book-recommendation
- Owner: philiptitus
- Created: 2025-03-21T12:15:28.000Z (2 months ago)
- Default Branch: main
- Last Pushed: 2025-03-21T12:16:56.000Z (2 months ago)
- Last Synced: 2025-03-21T13:27:25.647Z (2 months ago)
- Topics: adam-optimizer, bookrecommendsystem, collaborative-filtering, recommender-system, unsupervised-machine-learning
- Language: Jupyter Notebook
- Homepage:
- Size: 4.28 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Book Recommendation System using Collaborative Filtering
## Overview
This repository implements a collaborative filtering recommendation system for books using TensorFlow and the Goodreads dataset.
## Table of Contents
- [Features](#features)
- [Installation](#installation)
- [Technical Details](#technical-details)
- [Usage](#usage)
- [Model Architecture](#model-architecture)
- [Visualization](#visualization)
- [Contributing](#contributing)
- [License](#license)
## Features
- Matrix factorization-based collaborative filtering
- Handles sparse rating matrices
- Personalized book recommendations
- Interactive visualization tools
- Custom user ratings support
- Synthetic data generation for cold-start cases
## Installation
```bash
# Clone repository
git clone https://github.com/philiptitus/book-recommender.git
cd book-recommender
# Install dependencies
pip install numpy tensorflow pandas matplotlib seaborn scikit-learn
```
## Technical Details
### Data Structure
```python
# Rating Matrix (Y): User-book ratings (1-5)
# Binary Matrix (R): Rating indicators (0/1)
# Feature Matrices: W (users) and X (books)
num_features = 100
num_books = Y.shape[0]
num_users = Y.shape[1]
```
### Model Parameters
```python
W = tf.Variable(tf.random.normal((num_users, num_features)))
X = tf.Variable(tf.random.normal((num_books, num_features)))
b = tf.Variable(tf.random.normal((1, num_users)))
```
## Usage
### Data Preparation
```python
# Load dataset
books_df = pd.read_csv('books.csv')
Y, R = create_ratings_matrices(books_df)
# Initialize ratings
my_ratings = np.zeros(num_books)
```
## Model Architecture
### Collaborative Filtering Implementation
```python
def predict_ratings(X, W, b):
"""
X: Book features matrix
W: User features matrix
b: Bias terms
"""
return tf.matmul(X, tf.transpose(W)) + b
```
### Cost Function
```python
def compute_cost(X, W, b, Y, R, lambda_reg):
"""
Regularized cost function
"""
predictions = predict_ratings(X, W, b)
regularization = (lambda_reg/2) * (tf.reduce_sum(tf.square(X)) +
tf.reduce_sum(tf.square(W)))
return tf.reduce_mean(
tf.square(tf.multiply((predictions - Y), R))
) + regularization
```
## Visualization
### Learning Curves
```python
def plot_learning_curve(history):
plt.figure(figsize=(10, 6))
plt.plot(history)
plt.title('Model Learning Curve')
plt.xlabel('Iteration')
plt.ylabel('Cost')
plt.show()
```
### Rating Distribution
```python
def plot_ratings(Y):
plt.hist(Y[Y != 0].flatten(), bins=20)
plt.title('Rating Distribution')
plt.xlabel('Rating')
plt.ylabel('Count')
plt.show()
```
### Performance Metrics
```python
def evaluate_model(Y_true, Y_pred, R):
mask = R == 1
rmse = np.sqrt(np.mean(((Y_pred - Y_true)[mask]) ** 2))
mae = np.mean(np.abs((Y_pred - Y_true)[mask]))
return {'RMSE': rmse, 'MAE': mae}
```
## Contributing
1. Fork the repository
2. Create feature branch
3. Commit changes
4. Push to branch
5. Create Pull Request
## License
MIT License
```
Copyright (c) 2024
Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files.