Selecting a model
For this exercise, my goal is to simply show you how network data may be useful in ML, not to go into great detail about ML. There are many, many, many thick books on the subject. This is a book about how NLP and networks can be used together to understand the hidden strings that exist around us and the influence that they have on us. So, I am going to speed past the discussion on how different models work. For this exercise, we are going to use one very useful and powerful model that often works well enough. This model is called Random Forest.
Random Forest can take both numeric and categorical data as input. Our chosen features should work very well for this exercise. Random Forest is also easy to set up and experiment with, and it’s also very easy to learn what the model found most useful for predictions.
Other models would work. I attempted to use k-nearest neighbors and had nearly the same level of success, and I’m sure that Logistic regression...