Using PySpark to Create a Recommendation System for Amazon products.

PySpark recommendation system for Amazon products created for INFO 607 at Drexel University.

Project Overview:

The goal of our project was to create a recommendation system for Amazon user reviews on specific videos using PySpark. It was created as a final project for the class INFO 607: Applied Database Technologies at Drexel University. The data was downloaded from here. We focused on the Amazon Instant Video subset to optimize computation time.

File Manifest:

Folder /data - Contains all relevant data for the project.

• ratings_Amazon_Instant_Video.csv - Contains rating data for Amazon Instant Video products.
• video_names.csv - Contains names of videos, found manually (More information in Jupyter Notebook).
• ProductRecommender.ipynb - Main Notebook for this project.

Reason for Project

Recommendation systems are one of the go-to projects for learning how to work with PySpark and cloud computing. Creating recommendation systems can allow a company to predict what a user will or will not like and improving their experience accordingly.

Team Members:

Our team consisted of the following individuals (alphabetized by last name):

• Zach Carlson, zc378@drexel.edu
• Aviv Faraj, af3228@drexel.edu
• Katy Matulay, km3868@drexel.edu

Project Requirements

Because of the size of the dataset and how long training takes, we recommend running this on a Windows computer using a local configuration. See ProductRecommender.ipynb for more information.

Python Requirements

• Python ≥ 3.8.
• Python libraries required:
  • matplotlib.pylot
  • pandas
  • pyspark

How to Execute Notebook:

All of the code in this project needs to be opened in a Jupyter notebook environment. We recommend using Google Colab. See ProductRecommender.ipynb for more information.

Known Limitations of Project:

• Older dataset. The dataset contains ratings from 1999 to 2014. Having a dataset that includes more current ratings would help more accurately predict ratings. Additionally, in the last decade several genres have entered mainstream popularity such as LGBTQIA+ content, non-English speaking content, and limited series (e.g. Heartstopper, Squid Game, and Midnight Mass). Having ratings data on these more recent types of content can help improve accurately as well.
• Limited hardware. Training recomendation systems using Google Colab without paying for a premium account limits the size of data we can work with and the hyperparameters we test. Having a cloud platform account or better hardware would help with computation time and allow us to test more values before selecting a best model.