An R package for classifying Twitter accounts as bot or not.


Uses machine learning to classify Twitter accounts as bots or not bots. The default model is 93.53% accurate when classifying bots and 95.32% accurate when classifying non-bots. The fast model is 91.78% accurate when classifying bots and 92.61% accurate when classifying non-bots.

Overall, the default model is correct 93.8% of the time.

Overall, the fast model is correct 91.9% of the time.

API authorization

Users must be authorized in order to interact with Twitter’s API. To setup your machine to make authorized requests, you’ll either need to be signed into Twitter and working in an interactive session of R–the browser will open asking you to authorize the rtweet client (rstats2twitter)–or you’ll need to create an app (and have a developer account) and your own API token. The latter has the benefit of (a) having sufficient permissions for write-acess and DM (direct messages) read-access levels and (b) more stability if Twitter decides to shut down [@kearneymw](’s access to Twitter (I try to be very responsible these days, but Twitter isn’t always friendly to academic use cases). To create an app and your own Twitter token, see these instructions provided in the rtweet package.


There’s one function tweetbotornot() (technically there’s also botornot(), but it does the same exact thing). Give it a vector of screen names or user IDs and let it go to work.

fast = TRUE

The default [gradient boosted] model uses both users-level (bio, location, number of followers and friends, etc.) and tweets-level (number of hashtags, mentions, capital letters, etc. in a user’s most recent 100 tweets) data to estimate the probability that users are bots. For larger data sets, this method can be quite slow. Due to Twitter’s REST API rate limits, users are limited to only 180 estimates per every 15 minutes.

To maximize the number of estimates per 15 minutes (at the cost of being less accurate), use the fast = TRUE argument. This method uses only users-level data, which increases the maximum number of estimates per 15 minutes to 90,000! Due to losses in accuracy, this method should be used with caution!


In order to avoid confusion, the package was renamed from “botrnot” to “tweetbotornot” in June 2018. This package should not be confused with the botornot application.