Agile Data Science - How to Implement Agile Workflows For Analytics and Machine Learning

Sprints, Scrum, Kanban, Stories, Epics, Retrospectives, Extreme Programming, ’s opaque terminology and practices, plus the zeal of its advocates, can be off-putting to newcomers. Can it even be applied to data science, analytics and machine learning projects? In this talk we provide a gentle introduction to implementing an agile workflow for a data science team. We will demystify the terminology, tools and processes, and provide practical tips from our experience moving all of our client teams and projects to agile workflows in 2021. We’ve seen an increase in measurable output, better communication and a higher value-per-effort on work delivered. We’ve found it works especially well for managing research projects with a high level of uncertainty, such as developing machine learning models. Agile’s focus on measurable results aligns well with other goal-setting paradigms such as OKRs, but when applied to data scientific projects it encourages best practices such setting clear expectations on