Predictive Analytics using Machine Learning

Below you will read in the training and test data which are already split for you to load separately. Then use unnest() from tidytext to create the tidy version with one word per record. Now that you have train and test data loaded and tidied, you can see how many songs exist per artist/author. Since the dataset has songs and book pages, I'll refer to them each as a document. The features that you will create are based on documents and their associated metadata, so it's important to understand this concept.