Honours Thesis

Lowering barriers to using computational notebooks

by Tanaka Chitete

Abstract

Computational notebooks have revolutionised how data scientists, researchers, and educators engage with scientific computing—combining executable code, rich-text documentation, and visualisations. As their popularity has surged, so too has the realisation that the state-of-the-art solutions cause numerous frustrations, pain points, affecting a myriad of workflows. Chattopadhyay et al. (2020) taxonimised nine general workflows, indicating that users face numerous pain points across all nine, from setup to deployment. Grounded upon this work, we present Yuna, a novel self-contained computational notebook and platform that aims to sooth multiple identified pain points. We discuss Yuna’s implementation details, ranging from the interface implementation to session management. We thereafter evaluate Yuna with respect to the state-of-the-art, illuminating Yuna’s potential to address code management, reliability assurance, archival, dissemination, and reproducibility. Finally, we gather insights into current limitations and explore future directions. Yuna posits a step forward in the evolution of computational notebook platforms, democratising access and lowering barriers to their use.

Try Yuna here.