Use Claude Code to pull and chart public research data

An economist used Claude Code to gather Census homeownership data, clean it, chart it, and turn SEC filing text into a DuckDB research database.

Done with Claude Code· Intermediate· Oct 2026

What they did
In a webinar series, Paul Goldsmith-Pinkham asked Claude Code, a terminal-based assistant, to handle two research jobs. First, it tracked down Census homeownership data, worked through scraping issues, cleaned the spreadsheets, and wrote code to make the charts. Next, it pulled SEC EDGAR 10-K filings, isolated the risk-factor sections, and assembled them into a DuckDB database. He planned the work before each run and kept refining the output as needed.
How it went
The tariff-risk database showed tariff mentions becoming more common and more detailed in recent years. The source also said files that were messy at first became usable research data within a few minutes.
Tips
Start with a clear plan and expect to review and adjust the results as you go. For readers, it helps to ask for source links and a note on any gaps before accepting the final dataset or chart.

Try it yourself with Claude Code

I’m researching [topic]. Find the best public source for [data needed], download it, clean it, and save it as a DuckDB database in [folder]. Then make [chart type] and write a brief summary of what it shows, including source links and any caveats or missing pieces. If anything would be irreversible, stop and ask me to approve before continuing.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on economics.princeton.edu ↗

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.