Test the subreddit suggester on the newest 100 posts in each subreddit.

subreddit_suggester_test_new.py

import coremltools

# Scrape the 100 newest posts from each subreddit
max_posts = 100
posts = []
for subreddit in subreddits:
    posts.extend(get_n_posts(max_posts, subreddit, sort='new'))

# Apply the same preprocessing to the data
new_df = pandas.DataFrame(posts, columns=['subreddit', 'title'])
new_df.title = new_df.title.apply(lambda x: re.sub(r'\[.*\]', '', x))
new_df.title = new_df.title.apply(lambda x: re.sub(r'\W(?<![ ])', '', x))

# Load the mlmodel with coremltools for Python. Note you need to be
# on macOS for the predict function to work.
mlmodel = coremltools.models.MLModel('PATH/TO/subredditClassifier.mlmodel')

# Predict a subreddit for each title.
new_df['predicted'] = new_df.title.apply(lambda x: mlmodel.predict({'text': x})['label'])

# Mark the model correct if the predicted matches the actual subreddit
new_df['correct'] = new_df.predicted == new_df.subreddit

# Compute the fraction correct.
new_df.correct.sum() / new_df.shape[0]
# Output: 0.55
# We got 55% correct.