Ask questions of local CSV and parquet files in Python
Use an open-source library to ask plain-language questions of dataframes on your machine, with the generated code shown.
The use case
Analysts working in Python notebooks often want a quick answer without writing the pandas themselves. PandasAI lets them ask questions of dataframes, CSV or parquet files, and SQL sources; it generates and runs Python or SQL and returns results or charts. It can run with different model providers. The analyst still reads the generated code, checks results on a known case, and runs it in an environment where the generated code cannot reach sensitive systems. It suits quick checks during exploration; anything that feeds a report should be rewritten as plain, reviewed code.
Example workflow
- Install the library and set a model
- Load data into a dataframe
- Ask a question
- Read the generated code and result
- Copy confirmed code into the notebook
Tools
- PandasAI
- pandas
- Jupyter
Risks and guardrails
- Generated code runs locally; use a sandbox
- Data sent to hosted models may be sensitive
- Check results on a known example
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- PandasAI on GitHub: github.com/sinaptik-ai/pandas-ai