Recommendation-engine-IBM

Motivation:

The following project builds a recommendation engines with recommends articles for users on the IBM Watson Studio platform. It builds two forms of collaborative filtering recommendation engines:

• neighbour based
• model based using matrix factorization Effective recommendations enable users to more easily find content of interest

Getting Started:

In order to run the following code you will need to:

• Install all prerequisites below
• Run 'jupyter notebook' in the terminal and run the code in 'Recommendations_with_IBM.ipnb'

Prerequisites:

Data preparation

import pandas as pd import numpy as mp

Data visualisation

import matplotlib as plt

Files:

This repo contains the following folders and files

• Data - this folder contains data to build and test the recommendation engines

  • user-item-interactions.csv - a record for every interaction between a user and an article
  • articles_community.csv - information on the content in articles

• Recommenations_with_IBM.ipynb - a Jupyter notebook which holds the code to build and test recommendation engines

Tests and files to ensure the development of the recommendation engines is correct and bug free

• project_tests.py -
• top_5.p
• top_10.p
• top_20.p
• user_item_matrix.p