Udacity DSND Recommendations with IBM

Project is part of Udacity Data Science Nano-degree

Introduction

This project analyzes the interactions of users with articles on the IBM Watson Studio platform to make recommendations about new articles they will like.

Tasks

Exploratory Data Analysis

Rank Based Recommendations

Finds the most popular articles based on the most interactions. It is easy to assume the articles with the most interactions are the most popular. Rank-based recommendations can be helpful to recommend articles to new users.

User-User Based Collaborative Filtering

In order to build better personal recommendations for the users of the platform, similar user's interaction can be taken into consideration. These items could then be recommended to similar users. This will help generate personal recommendations for the users.

Matrix Factorization

Create matrix decomposition using user-item interactions. With it, new article interaction can be predicted to an extent.

Discuss possible methods for moving forward and test recommendations.

Files

1. Recommendations_with_IBM.ipynb - Code and analysis
2. data - dataset folder
3. project_tests.py - tests scripts