Photo by Charles Deluvio on Unsplash

Recommendation System App

Check the web APP

Description

Usually, when I go on Youtube I like to find videos related to Data Science or Deep Learning that I have not seen before. However, sporadically I like to find songs, humoristic videos or even watch some matches of professional League of Legends. The problem with this is that Youtube would start recommending me music or videogames clips in which I am not really interested.

Then, one night in bed I had a great idea “What if I could create a recommendation system, with different clusters and only see recommendations of the cluster I am interested in”.

Next day, I checked on Youtube and that option was already implemented, but since I don’t use the platform that much I had not been aware of it.

Anyway, I wanted to test my idea, so I decided to build that system myself. As a proof of concept, instead of using the Youtube database, I will implement it using the MovieLens Dataset, that can be downloaded from Kaggle.

NOTE: Yes, I know! If you visit the projects at Kaggle there are hundreds of repeated implementations of the same EDA processes and the same methods for recommendations using this dataset. This project aims to be more unique and personalized implementing other methods and a web user interface using Streamlit

App interface

(CURRENTLY IT WORKS LOCALLY, WORKING ON DEPLOYMENT)

Components

This project involves:

  • Crawling the web and parsing data
  • Interacting (pulling and inserting data) with a database
  • Machine Learning algorithms
  • Personalized widgets

For more details visit the GitHub page associated to this project.

Jason Omedes
Ph.D. Brain Machine Interfaces

My research interests include artificial intelligence, neuroscience, robotics and everything related to finding patterns and solving the mysteries hidden in data.