Title

Improving Data Science Team Performance via the Use of a Kanban Process Framework that has Enhanced Mentoring, Coaching and Metrics Utilization

Date of Award

December 2019

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

School of Information Studies

Advisor(s)

Jeffrey S. Saltz

Keywords

Data Science Project Management, Kanban, Kanban Coach, Kanban Master, Process Methodology, Project Management

Subject Categories

Social and Behavioral Sciences

Abstract

The amount of data generated by organizations and systems is growing exponentially. At the same time, organizations are increasing their efforts to generate knowledge and useful insights that might be hidden within their data. This is driving tremendous growth in teams doing data science and the related field of big data. However, despite the fact that data science is now widely used in many organizations, most data science teams do not follow any specific methodology of how to organize, collaborate and structure the team’s work.

Kanban is one of the potential methodologies that can be used to help a data science team improve their coordination and effectiveness. The use of Kanban is increasing across a range of information system projects, including software development and data science. While the use of Kanban is growing, little has been done to explore how to improve team performance for teams that use Kanban. One possibility is to introduce a process coach. Another possibility is to introduce a process master. Yet another possibility is to collect Kanban team metrics (to be able to understand and predict low team performance).

This paper reports on an experiment comparing teams using a Kanban Coach (KC), a Kanban Master (KM) or neither a KC nor a KM. Coordination Theory and Shared Mental Models were employed to provide an explanation as to why a KC or a KM might lead to better

project results. It was found that introducing KC or a KM led to significant improvement of team performance in comparison to the baseline case, with the KC being better than the KM with respect to how teams used Kanban. This dissertation also explored, for teams using Kanban, the ability to predict low team performance via an analytical model that uses specific project metrics that can be collected via the team’s visual Kanban board. The models developed, via the analysis of 80 data science teams using Kanban, were significantly better than the baseline situation of thinking that all teams were at risk.

Finally, note that while this research was done within a data science project context, the results are likely applicable across a range of information system projects that use Kanban.

Access

SURFACE provides description only. Full text may be available to ProQuest subscribers. Please ask your Librarian for assistance.

This document is currently not available here.

Share

COinS