Python and R for the Modern Data Scientist

Chapter 5. Workflow Context

Boyan Angelov

A common source of frustration for data scientists is discussing their work with colleagues from adjacent fields. Let’s take the example of someone who has been working primarily in developing ML models, having a chat about their work with a colleague from the business intelligence (BI) team, which is more focused on reporting. More often than not, such a discussion can make both parties uncomfortable due to a perceived lack of knowledge about each other’s work domain (and associated workflows)—despite sharing the same job title. The ML person might wonder what D3.js is, the grammar of graphics, and all that. On the other hand, the BI data scientist might feel insecure about not knowing how to build a deployable API. The feelings that might arise from such a situation have been termed impostor syndrome, where doubts about your competency arise. Such a situation is a by-product of the sheer volume of possible applications of data science. A single person is rarely familiar to the same extent with more than several subfields. Flexibility is still often required in this fast-evolving field.

This complexity sets the foundation for the workflow focus in this chapter. We’ll cover the primary data science workflows and how the languages’ different ecosystems support them. Much like Chapter 4, at the end of this chapter, you’ll have everything needed for making educated decisions regarding your workflows.

Defining Workflows

Let’s take a step ...