We are introducing first-class Jupyter Notebooks support in Datagrok. In addition to the exploratory data analysis capabilities of the Jupyter platform that were available as part of Datagrok for a while, we now support the following data analysis pipeline:
- open table in a notebook
- transform and visualize using any available in Jupyter Notebooks functionality
- return result into Datagrok
Integration is done as an open-source Notebooks package for the Datagrok platform.
Vasiliy, this is a great feature! I already have another idea Why don’t we provide a capability to automatically extract a (script-based) function out of the notebook? Here’s the scenario I have in mind:
- The user opens a table in notebook
- The user writes a series of steps in the notebook, arriving to the result (last cell’s value)
- The user clicks on the “Save as function” link, at which point we generate the source code for a parameterized Datagrok Python function. Its input is the data frame, and the output is the last cell’s value. This generated code gets shown in the “New script” view, so the user can adjust it (modify parameters, etc) before saving it.
This feature might dramatically simplify the process of developing new scripts, since it’s a lot easier to debug them in the Jupyter environment.