The Whole is More Than the Sum of Its Parts: Coordinated Visualizations for Code Understanding and Error

Abstract

Visualizations have the potential to help non-coders and coders alike understand and work with any programming language across a variety of tasks. This is especially salient in code for data visualization tasks where errors may lie not just in the code itself, but in the many potential data transformation and visual design choices. To help navigate this type of work, we review existing code aids and focus in on four main methods based on an iterative design process. These four designs were then mixed and tested by our team, resulting in a hybrid approach to help users interpret data analysis code. This programming language-agnostic hybrid visual design was developed to translate data transformations and chart building code for users, capturing what each line of code accomplished with respect to the data. This work has the potential to improve programming education, code evaluation and verification, and data visualization processes.