Analyses
Introduction
A Mantle analysis object encapsulates an analysis notebook and links to the notebook’s inputs, as well as its output datasets and files.
Getting an analysis by unique ID
Adding inputs to analyses
Adding a Mantle dataset as an input to an analysis
Mantle datasets can be retrieved using their unique ID and added as inputs to an analysis.
Adding a file input to an analysis
Track file inputs in your analysis by adding file inputs, specifying a key to identify the input and the file’s local path.
Adding a Python string or Boolean input to an analysis
Track string or Boolean inputs by specifying the input key and value.
Adding outputs to analyses
Adding a dataset output to an analysis
Add dataset outputs to an analysis to track the generated outputs, providing a key to identify the output as well as the output file’s path.
Adding a file output to an analysis
Add file outputs to an analysis to track the generated outputs, providing a key to identify the output as well as the output file’s path.
Adding a Python string or Boolean output to an analysis
Track string or Boolean outputs by specifying the output key and value.
Advanced
Advanced
Creating a Mantle analysis outside the Mantle UI
When you use analysis notebooks within the Mantle user interface in combination with the Mantle SDK, you can track and share the executed code in addition to cataloging the inputs and outputs.
You can also create an analysis object outside of Mantle analysis notebooks using the SDK. Some examples of this use case include analysis performed within local Jupyter notebooks, local Python cripts, or a Python interpreter. However, though the analysis and its inputs and outputs will be represented in the lineage graph in the Mantle UI, the code will not be tracked or accesible within the Mantle UI.
Creating an analysis object
Note: Local analysis creation does not currently support tracking of notebook code or any other code executed outside of the Mantle UI. For full visibility into the executed code version, using the Mantle UI is recommended.