dorothearos / IBM-Recommendations

Recommendation Engine using real data from IBM Watson Studio platform.

Recommendations with IBM

Table of Contents

1. Description
2. Getting Started
   1. Dependencies
   2. Installing
   3. Additional Material
3. Authors
4. License
5. Acknowledgement

Description

This project analyzes the interactions that users have with articles on the IBM Watson Studio platform. The project aims to make recommendations for users on the IBM platform about new articles that they might like.

The Project is divided in the following Sections:

1. Exploratory Data Analysis
2. Rank Based Recommendations
3. User-User Based Collaborative Filtering
4. Content Based Recommendations
5. Matrix Factorization

Getting Started

Dependencies

• Python 3.5+ (I used Python 3.8)
• Machine Learning libraries: NumPy, SciPy, Pandas, Scikit-Learn
• Natural Language Process libraries: NLTK

Installing

Clone this GIT repository:

git clone https://github.com/lng15/IBM-Recommendations

Additional Material

data folder contains dataset which fetched from IBM Watson Studio:

• articles_community.csv dataset contains articles information
• user-item-interactions.csv dataset contains users information, as well as users-articles interactions.

There are two notebooks in this repo:

1. Recommendations_with_IBM.ipynb : the main notebook which contains all Sections mentioned in the Description.
2. Content Based.ipynb : the extra notebook which is made as another way to recommend content to users. It is not required to finish the project and it will be improved in the future. However, its result can be used to compare to the approach in the main notebook above.

Authors

• LN

License

This project is licensed under the MIT License - see the LICENSE.md file for details

Acknowledgements

• Udacity for giving me a chance to work on the project
• IBM for providing the permission to use data on the platform