Databot: AI-assisted data analysis in R or Python

If you’d like an LLM to act more like a partner than a tool, Databot is an experimental alternative to querychat that also works in both R and Python. Databot is designed to analyze data you’ve imported into your working session and suggest questions you might want to ask. After you decide what to ask, Databot will write the code to generate your answer. Posit CTO Joe Cheng, creator of querychat, also wrote this package. It’s now available as an add-on for the Positron IDE.

Also see the introduction to querychat that accompanies this sidebar.

Databot is “an experimental AI assistant that is designed to come alongside you, and help you by performing tasks within your R session,” according to the R documentation. “Point Databot at some data and it will come up with plenty of ideas of how to analyze it.” It currently only uses Anthropic’s Claude Sonnet 3.5, so you’ll need an Anthropic API key to use it.

Note that when the docs say “experimental,” they mean it. “Pretty fragile currently,” the README warns.

With that in mind, I installed databot with remotes::install_github('jcheng5/databot'), loaded my NFL games data set and data dictionary file into my R session along with the package, and created the following chat:

games