Google Gemini Python data skill
Run complex calculations, regressions, and aggregations using pandas and scikit-learn in the sandbox.
Workflow
- Write a Python script using
pandas,numpy, orsklearnto:- Join, filter, and aggregate tables.
- Compute statistics (mean, median, stddev, correlations).
- Build or evaluate ML models.
- Run statistical tests (chi-square, t-test) when appropriate.
- Execute in the sandbox. Run via
python3 script.py. - Handle errors. If the code fails, read the traceback and fix it.
- Return structured output. Print results as JSON for downstream skills.
Example output
[
{"segment": "High Risk", "churn_rate": 0.34, "avg_tenure": 2.1},
{"segment": "Low Risk", "churn_rate": 0.05, "avg_tenure": 14.7}
]
Notes
- Use
select_dtypes(include=["object", "str"])for categorical columns (avoids the pandas 4 deprecation warning). - Install missing packages with
pip install -q <package>before importing.