A training dataset is a collection of examples with known categories, or labels, used to help a machine-learning system learn patterns and make predictions. section 11.1
A supervised-learning training dataset contains:
- Objects to classify, such as book texts. section 11.1
- Known categories or labels, such as the authors of those books. section 11.1
- Features calculated from each object, such as average word length or average words per sentence. section 11.1 section 11.8
Identifying an unknown author
Train on books whose authors are known, calculate writing-style features for each book, and use the learned patterns to predict the author of a mystery book. section 11.1
