Problems with quantitative categorization: An argument for qualitative approaches

Abstract

As data science gains traction, it often brings quantitative approaches and positivist epistemologies. While these can generate powerful insights, we argue for methodological hybridity in modern data science. We demonstrate the power of complementary qualitative approaches and flexible ontologies. Using an example of classifying segments™ on Strava, neither quantitative nor qualitative approaches alone were adequate to meaningfully classify segments, but together allowed accurate, useful, and intuitive categories to emerge. Drawing on this experience, we discuss qualitative data science and argue the ontological discussions within Critical GIS from the 1990s and 2000s are increasingly relevant and informative amidst our platial paradigms.