Recommendations-with-IBM

Table of Contents

1. Project Motivation
2. Summary of Results
3. Instructions
4. Libraries Used
5. File Descriptions
6. Acknowledgements

Project Motivation

Recommendations with IBM

For this project, we will analyze the interactions users have with articles on the IBM Watson Studio platorm and provide recommendations on which new articles you think they will like. To determine which articles to show to each user, we will be performing a study of the data available on the IBM Watson Studio platform.

The project will be divided into the following parts:

I. Exploratory Data Analysis

Before making recommendations, we will did to explore the data that we are working with for the project. We'll answer the basic required questions in the data that that we'll be working on throughout the rest of the notebook before we dive into the details of the recommendation system in the later sections.

II. Rank Based Recommendations

To begin building recommendations, we'll need to find the most popular articales simply based on the most interactions. Since there are no ratings for the articles, we can easily assume the articles with the most interactions are the most popular. These will be the articles we might recommend to new suers (or anyone depending on what we know about them.

III. User-User Based Collaborative Filtering

To build better recommendations for the users of IBM's platform, we could look at users that are similar in terms of the items they have interacted with. The items could then be recommended to the similar users. This would be a step in the right direction towards more personal recommendations for the users. This step will be implemented next.

IV. Matrix Factorization

We will complete machine learning approach to building recommendations. The user-item intereactions will build out a matrix decomposition where we'll gain an idea of how well we can predict new articles an individual might intereact with. We will also discuss which methods we may use moving forward and how well our recommendations are working for engaging users.

Summary of Results

Analyzed the user articles from articles_community.csv and the user interactions from user-item-interactions.csv and producing different recommendation engines and analyse their performance.

Web app:

The web app can be accessed here: https://recommendationibm.herokuapp.com/

The web server run the recommendation by the following steps:

• The web server invokes recommendation_app.py via Procfile
• recommendation_app.py invokes rec_app.init.py and listen on 0.0.0.0:3001
• rec_app.init.py invokes rec_app.run.py to response web request.

Instructions

Clone this repo to your computer

Ensure jupyter is installed correctly using jupyter --version and this should return jupyter version

cd to the location of the project home directory

Execute Recommendations_with_IBM.ipynb

Libraries and Dependencies

Python 3.6.6+

1. pandas
2. numpy
3. matplotlib
4. pickle
5. scikit-learn

File Descriptions

Recommendations_with_IBM.ipynb: Jupyter notebook containing main implementation and analysis.

top_5.p: pickle file to test top 5 rank based recommendations

top_10.p: pickle file to test top 10 rank based recommendations

top_20.p: pickle file to test top 20 rank based recommendations

user_item_matrix.p: pickle file to load the user_item matrix

project_tests.py: contains the unit tests for the solution It is recommended you run the solution from a jupyter notebook.

web_app:

• web app files including procfile and requirements.txt for heroku

data:

• articles_community.csv: csv file containing the articles

• user-item-interactions.csv: csv file containing user interactions with articles

Acknowledgements

1. https://www.ibm.com/ca-en
2. https://www.udacity.com/