Book Recommendations with ALS on PySpark

This repository contains an implementation of the Alternating Least Squares (ALS) algorithm using PySpark on a Google Cloud services. The project focuses on generating collaborative book recommendations based on user ratings data .

Repository Description

The project involves the following steps:

1. Loading book ratings data from a CSV file stored in a GCS bucket.
2. Storing the loaded data into a BigQuery table using a PySpark cluster.
3. Training an ALS model on the book ratings data to generate recommendations.
4. Evaluating the model using the Root Mean Squared Error (RMSE) metric.

Files

csv_to_bigquery.ipynb

This notebook performs the following tasks:

• Loads a CSV file containing book ratings data from a GCS bucket.
• Uses a PySpark cluster to process the data.
• Stores the processed data into a BigQuery table.

als_recommendation.ipynb

This notebook performs the following tasks:

• Initiates a Spark session with BigQuery configuration.
• Loads data from the previously created BigQuery table into a DataFrame.
• Uses the ALS algorithm to train a recommendation model on the book ratings data.
• Calculates the RMSE score to evaluate the model's performance.

Setup and Usage

Prerequisites

• Google Cloud Platform (GCP) account
• Google Cloud Storage (GCS) bucket
• Google BigQuery
• Google Cloud SDK
• PySpark cluster created on GCP

Steps

1. Clone the Repository

   git clone https://github.com/zygmuntz/goodbooks-10k.git
   

2. Upload the CSV file to GCS

   Upload the ratings.csv file to a GCS bucket.

3. Configure Google Cloud SDK

   Install google cloud SDK and Authenticate your GCP account:

   gcloud auth login
   gcloud config set project your-gcp-project-id

4. Run csv_to_bigquery.ipynb

   Open the notebook on pyspark cluster and follow the instructions to load the CSV file from GCS and store the data in a BigQuery table.

5. Run als_recommendation.ipynb

   Open the notebook on pyspark cluster and follow the instructions to initiate a Spark session, load data from BigQuery, train the ALS model, and evaluate its performance.