Following the User's Footsteps: A Data-driven Framework for Measuring and Modeling Elegant Design in Engineering
Abstract
Elegance and elegant design is a widely sought property in different disciplines, especially in engineering where solutions to design problems have many different factors that influence their efficiency and usability. However, the lack of literature on this construct and consensus on the definition of elegance limits our understanding of its role and applications. The present research study applies various qualitative and quantitative statistical methods to provide a valid and reliable scale for measuring elegant design in engineering. The measures of elegant design are developed from a comprehensive review of the literature and filtered by expert judges' ratings. The resulting elegant design scale is validated through exploratory and confirmatory analysis and tested by comparing the scale's results against subjective assessments. Results from this research study are further applied to build prescriptive and predictive models of elegance using generalized linear regression, K nearest neighbors and random forests methods.. This study proposes a scale for defining and measuring elegant design, which consists of thirteen items belonging to three latent categories: functionality, simplicity and aesthetics. Furthermore, it introduces classification models of elegant design constructed from measures, subsets of measures, and factors of elegance that distinguish elegant designs from non-elegant ones. Current results show that measures of elegant design can be used to achieve various levels of prediction accuracy for classification depending on the targeted complexity and interpretability. Accurate and reliable tools that enable decision makers to evaluate design solutions in engineering in terms of frequently overlooked concepts like user engagement are of significant value. The current research is an attempt to introduce a tool of such nature and paves the path toward the development of more.