MenuSights is currently two things:
A "real-world" dataset being used to explore and illustrate the utility of deep neural networks and word vector embeddings in extracting cholesterol predictions from the names of food items. This is under development in this Jupyter notebook.
A webtool developed during my time with Insight Health Data Science. It's based on a logistic regression classifier that analyzes names of items on local restaurant menus and provides a prediction as to whether the food is likely to be low, medium, high, or very high in cholesterol. There is a much fuller explanation of how the data were collected and how the model was trained evaluated within the relevant Jupyter notebooks