Overview

There are multiple drugs available for a condition, and consumers often have difficulties choosing drugs for their conditions.
My aim of this project is to make a recommendation system for patients by predicting patients outcome using drug reviews.

The original dataset is from UCI Machine Learning Repository.
https://archive.ics.uci.edu/ml/datasets/Drug+Review+Dataset+%28Drugs.com%29

The recommender system potentially can help patients to choose better drugs for their conditions, and also can provide benchmark to drug providers such as doctors and pharmaceutical companies.

How to Run

1. Getting dataset: https://archive.ics.uci.edu/ml/datasets/Drug+Review+Dataset+%28Drugs.com%29
2. Data cleaning and initial EDA: drug_recommendation-data_cleaning.ipynb
3. Topic modeling: drug_recommendation-topic_modeling-CV_bigram-LDA.ipynb
4. Distribution of topics of top 10 conditions: drug_recommendation-top_10-topics.ipynb
5. Supervised learning: drug_recommendation-nlp_supervised.ipynb
6. Summary of my data: presentation slides drug_review.pdf