Build Classifier from Google Query dataset

I want to build a classifier based on the publicly available Reddit comment database on BigQuery. What this involves is running a query to get the most used words in each subreddit, then saving those most-used words to file (without stopwords), and using the files to build a simple sklearn Bayes / SVM classifier using sklearn in Python. I would do this myself but I'm pressed for time on other things. This allows us to classify text based on subreddits. More details available on request. Only serious people please. Thanks!