Context For a [Springboard](https://www.springboard.com/) project. Potentially part of a larger project to infer demographic information based on the text that people write. And hopefully generate some sociolinguistic insights in the process. Some of the code associated with this project (including most of the code used to generate these data) is in [this Github repository](https://github.com/andyharless/twit_demog), but it is still a bit of a mess. Content The files contain the text of tweets downloaded via the Twitter API between 2019-05-21 and 2019-06-01, with a user ID and timestamp to identify each tweet, and an indicator of whether the user appeared (based on the display name) to be male or female. The tweets are limited to English, original tweets (i.e., not retweets) associated with users whose first names could be identified by the [gender-guesser ](https://pypi.org/project/gender-guesser/) package as male or female. (Names are not included.) Tweets are divided into train, validation, and test sets by time and user ID. (In other words, there should be no overlap in time or user ID between the files, and the sequence from train to validation to test should be strictly forward in time.) The tweets in each file have been selected to be evenly balanced between those identified as male and female (which usually means keeping all the "female" tweets and selecting a random subset of the "male" ones. Tweets have been preprocessed (using [a script](https://gist.github.com/timothyrenner/dd487b9fd8081530509c) copied from Timothy Renner) to remove hashtags, mentions, URLs, media, and symbols. Acknowledgements I thank Twitter, Kaggle, the authors and maintainers of the gender-guesser package, the users who wrote the tweets, and all the various open source people who work on the packages that make data munging possible for those who can't afford fancy software. Inspiration How well can we train a machine learning model to guess the gender of Twitter users based on the text of their tweets? And what can such models tell us about the gender sociolinguistics of Twitter users?