Data 643 - Final Proposal

July 9, 2017

Ann Liu-Ferrara

Objective

I would like to build a recommender system with information other than user id and item id, like gender or images.

Dataset

I found the dataset MovieLens is very interesting. The website https://movielens.org/ apply different recommender systems to users with different rating status, i.e. before and after rating 15 movies; also the dataset contains user demographic information, such as gender, occupation, and image.

http://files.grouplens.org/datasets/movielens/ml-100k.zip

Methodology

I am planning to use deep learning method in Python Jupyter notebook to implement the recommender system. This will be my first time to use Python Jupyter notebook for project deliverable, and it will be a learning curve. I am not sure how deep I would like to get involved in deep learning in this project.

Resources

http://blog.nycdatascience.com/student-works/deep-learning-meets-recommendation-systems/

http://blog.ethanrosenthal.com/2015/11/02/intro-to-collaborative-filtering/

Deliverables

I will publish code in Python Jupyter Notebook to GitHub and rPubs.