# define the training set
train_data = fetch_20newsgroups(subset="train", categories=text_categories)