Jupyter notebooks are rapidly becoming the default way to apply data science. They can be seen as Google Docs for analytics. Documentation, lines of code, and output from that code (text, data tables, charts, and plots) are all united.
In addition, such notebooks are easily shareable, even without a backend. Just export them as JSON documents.
Jupyter notebooks preserve all output cells even if the cell during execution has been connected on a large scale Apache Spark cluster processing hundreds of gigabytes of data.
In my experience, notebooks are used mostly in the following scenarios:
- Notebooks are an ideal tool for creating and sharing a knowledge base on best practices a core data science team executes. This way, their knowledge is documented in an executable way.
- Notebooks are used to document the variety of available data sources in an...
