Jupyter Notebooks

The NotebookContent block embeds a full Jupyter Notebook right into a report — next to the charts and tables. Complex statistics, forecasts and ML models live in the same place as the business report.

Execution

• the notebook runs through papermill + ipykernel;
• the result is displayed next to the report’s other blocks;
• the execution timeout is 10 minutes;
• the block shows the status and the time of the last save.

The automatic environment

• Python 3 is already built into the Docker image — nothing to install on the host;
• the service creates the virtualenv itself on the notebook’s first run;
• dependencies are installed from data/requirements.txt — the file is seeded on first launch, and you can extend it with your own libraries.

Built-in libraries

• pandas and NumPy — data processing;
• Matplotlib and Seaborn — visualization;
• SQLAlchemy — access to databases.

Templates and storage

• notebook templates: basic, data-analysis, machine-learning;
• two storage options: the notebook’s JSON embedded in the block, or an external .ipynb file;
• the block displays a cell counter.

Access to data

Through SQLAlchemy the notebook connects to your databases, and Lakehouse files (Parquet, SQLite, Markdown) are read directly from the data folder. Details in the Lakehouse section.

← Back to the documentation index