Medical Service Rating Machine Learning

Outcome

This project aims to use a large dataset of medical service providers’ reviews and ratings to predict doctors’ ratings accurately.

I used text tokenization, sentiment analysis, and categorization variables to build a machine-learning model for the doctor’s ratings. The model is cross-validated with a 90.8% accuracy. By examining the keywords that contribute the most to the rating we can find the most important words that matter to the patient’s opinion.