Deconfounded Causal Inference

Machine Learning Term Project with Prof. Rajesh Ranganath at NYU. Spring '22

This project aims to study the effectiveness of using Deconfounded Causal Inference for Recommendation.

Setup

Add any depedencies+ environments here

Jobs

• Preprocessing of the raw text files not required, use the Jobs Notebook to read directly from the data

Books

• CLEANED data ready for use is in the data folder. Preprocessing code in Books-Preprocessing

Data Links

• Jobs Data
• Books Data
• Movies Data

Results

• Results Spreadsheet

Useful Links

• Deconfounder Paper
• Deconfounded Recommendation Paper.
• Deconfounded Code Tutorial
• Deconfounded Causal Inference Slides
• Deconfounder Code
• Poisson Factorization Code
• Hirarchical Poisson Factorization Code