Creating Analysis Notebooks with AI

Creating Analysis Notebooks with AI#

In this guide, you’ll use AI-powered Copilot Chat to automatically generate Jupyter notebooks for measurement analysis. This is ideal for creating reproducible workflows like thermal analysis, crash analysis, or performance comparisons.

Overview#

This playground demonstrates:

  • Using AI to design analysis workflows

  • Generating complete Jupyter notebooks

  • Running analysis and generating visualizations

  • Saving notebooks for reproducibility and sharing

Step 1: Define Your Analysis Goal#

First, determine what analysis you want to perform. Some examples:

  • Thermal Analysis — Compare temperature measurements across test campaigns

  • Performance Comparison — Analyze acceleration or efficiency metrics

  • Data Quality Check — Identify anomalies or missing values

  • Correlation Analysis — Find relationships between different measurements

Step 2: Ask Copilot to Create a Notebook#

In Copilot Chat, describe your analysis goal:

I want to create a thermal analysis of measurements from test campaign 'Campaign_01'. 
Show temperature trends and identify peak values.

Copilot will:

  1. Ask clarifying questions (e.g., date range, temperature thresholds)

  2. Generate a complete Jupyter notebook with:

    • Connection setup code

    • JAQueL queries to retrieve data

    • Data cleaning and preprocessing

    • Analysis and visualization code

    • Statistical summaries

Step 3: Review and refine the Generated Notebook#

If the generated notebook needs adjustments, provide feedback:

Add a second plot comparing temperature from two different test campaigns

Include error handling for cases where data is missing

Add a summary table with statistics for each measurement

Copilot will update the notebook accordingly.

Tips & Best Practices#

  1. Be descriptive — Provide context about your data and analysis goals

  2. Start simple — Begin with basic analysis, then add complexity

  3. Verify queries — Have Copilot explain generated JAQueL queries before execution

  4. Test on small data — Use $rowlimit to test queries on a subset first

  5. Document assumptions — Add markdown cells explaining data filters and calculations

  6. Version control — Commit notebooks to Git for tracking changes

Next Steps#

Next: Creating Analysis Applications