Towards Scaffolding Complex Exploratory Data Science Programming Practices

Abstract

Although a wide range of professional and end-user programmers want to engage today with data science programming, this form of programming presents unique challenges. For instance, data science tasks typically require exploratory iterations: coding and running many different approaches to reach a desired result [1]–[3]. In a body of research building towards my thesis, I have interleaved behavioral studies of data scientists with systems building research towards scaffolding new forms of support for keeping track of iterations during this experiment-driven form of work.