Paper: Notebooks for Data Analysis and Visualization
Computational notebooks offer an alternative to the common GUI-based tools used for data visualization and BI today. In this new paper, I talk about what they are, their pros and cons, and how research could fill in some important gaps.
Data visualization research has focused primarily on graphical user interfaces (GUIs) for creating data visualization, and for good reason. But notebooks have been used in data science for a while now, and they offer their own advantages over GUIs: reusability, integration of data analysis and modeling, and – especially – easy collaboration.
This is an invited piece for the Graphically Speaking column in CG&A, and I'm obviously biased because I work for Observable now. There are a fair number of computational notebook platforms out there though, like R Markdown in RStudio, Jupyter for Python, etc.
The paper talks about what notebooks are, where I see their strengths (and some weaknesses!), and in particular where I see opportunities for research. What I don't discuss in the paper, since it wasn't as big a hype when I wrote it as it is now, is that notebooks are also pretty ideal for exploring the current wave of AI tools, in particular ChatGPT and similar. But whether it's old-fashioned data analysis and visualization, financial or other modeling combined with analysis, or exploring AI models, I think there's a large research space here that is largely untapped.
Robert Kosara, Notebooks for Data Analysis and Visualization: Moving Beyond the Data, Computer Graphics & Applications (CG&A), vol. 43, no. 1, pp. 91-96, 2023.