Book Recommender System

This repository contains a series of Jupyter notebooks and datasets that document the process of creating a book recommender system. The system was built using data scraped from Goodreads, covering the top 100 best books published in 2023, an extensive GoodReads dataset containing 100k books and utilizes a KMeans clustering algorithm to provide recommendations.

Repository Structure

• 0. Data Links and Explanations: This notebook contains links to the data sources used, along with explanations of why and how the data was gathered.

• 1. Data Import and Inspection: The initial data import steps are detailed here, along with the first look at the raw data structures.

• 2. Data Cleaning: A crucial step where the data is cleaned and preprocessed, including handling missing values, outliers, and standardizing text values.

• 3. Exploratory Data Analysis (EDA): This notebook explores the data visually and statistically to uncover patterns, trends, and relationships.

• 4. Feature Engineering and Model Preparation: This stage refines the dataset by performing sentiment analysis on book descriptions, encoding genres, and normalizing numerical features. The process culminates in the use of K-Means clustering to group books, with the final model evaluated based on cluster homogeneity and the silhouette score, eschewing 'total_ratings' for a content-based recommendation approach.

• 5. Book Recommender: The final notebook where the recommender system is implemented, tested, and validated.

• Raw Data: This directory contains the initial datasets as they were collected, without any modifications or cleaning. These are the datasets directly extracted from their sources.

• Clean Data: Here, you'll find the cleaned datasets ready for analysis, including the 'goodreads_webscrap.csv' file which contains my data scraped from Goodreads.

• Pickle: This folder houses the serialized Python objects: the trained K-Means clustering model ('kmeans.pkl'), the scaler ('scaler.pkl'), and the encoder ('encoder.pkl'), which are essential components of the book recommendation system.

• Presentation: Contains the final presentation in PDF format, along with a link in the "Link Tableau Dashboard" file to Tableau visualizations that offer a graphical representation of some project's insights and outcomes.

---

How to Use

To use the recommender system:

1. Ensure all files from the Data folder are downloaded. Make sure to run everything in the same directory.

2. Start with notebook 0. Data Links and Explanations to understand the data sources.

3. Proceed through the notebooks in numerical order, as each notebook builds upon the work done in the previous ones.

4. To get a recommendation, run the final notebook 5. Book Recommender. You will be prompted to enter a book title and author name. Based on this input, the system will either provide a recommendation from the existing dataset or fetch information online to generate a recommendation.

5. ⚠️ NOTE: Due to size constraints on GitHub, not all CSV files are uploaded. To utilize the book recommender, download 'final_df.csv' and 'goodreads_df_cleaned.csv' from the Clean Data folder, as well as 'kmeans.pkl', 'scaler.pkl', and 'encoder.pkl' from the Pickle folder. Run the "5. Book Recommender" notebook in the same directory for immediate use without the need for prior execution of other notebooks.

Documentation

Each notebook is well-documented with markdown cells explaining the steps taken, the rationale behind them, and interpretations of the results. Comments in code cells provide additional context for the code being run.

Contributing

Contributions are welcome, and any feedback or suggestions can be made by opening an issue or a pull request.

---