A simple experiment to enrich twitter profiles with gender

1. Clone Repository

2. Install Requirements: pip install -r requirements.txt

3. Run the file processor python file_processor.py

—> It automatically downloads the dataset, builds the classifier, prints the test results and stores an enriched version of the dataset (csv file) under: twitter_gender/data/users_enriched.csv

The last two columns are added:

• ‘prediction’: the prediction result ‘female’ or ‘male'
• ‘source’: If the final prediction has happened based on the name (’name) or on description + last tweet (’text’)

Yes.

from twitter_gender.file_processor import ProcessUsers

process_users = ProcessUsers()

labels = ['unknown', 'female', 'male']

gender = process_users.gender_classifier.get_gender_by_name('Johannes Erett')
print(labels[gender])
  >> 'male'

gender = process_users.gender_classifier.get_gender_by_text_custom('I am a woman of great faith... in unicorns ❤️ ')
print(labels[gender])
  >> 'female'

• Currently the classifier is not stored, but alsways built and trained from scratch. I would recommend to pickle and re-load it.

• Code needs better structure and more documentation.

Tested with Python 3.7.1