In this Jupyter Notebook project, a suspected correlation is explored between the text content of a business' user reviews and the business' respective star rating. By applying natural language processing (NLP) to the text content of user reviews, a supervised machine learning model in TensorFlow can be constructed to predict star ratings.
The motivations for this project to uncover a suspected correlation using machine learning; ergo, the built model here is not necessarily optimal. If a business has enough reviews, there is a surprisingly accurate prediction and correlation between the text content of the user reviews and the star rating a business receives.
While GitHub supports rendering Jupyter Notebook projects, it does not support rendering them as default README page. Furthermore, there are performance issues when rendering non-trivial notebooks. For best results, the notebook can be viewed using Jupyter's nbviewer.
The dataset used for this project is sourced from the Yelp Open Dataset, a public and non-commercial dataset for educational purposes.