Machine Learning Project Management - A Study of Project Requirements and Processes in Early Adoption
Martin Lindovsky, Christoffer Brasjö · Chalmers Publication Library (Chalmers University of Technology) · 2019
Machine learning projects have increased in numbers and appropriate project management processes and methodologies are needed to ensure project success.Insufficient research has been done on the subject which could support corporations and organisations that are planning to start, or have recently started, with applied machine learning.The purpose of this thesis is to provide project management guidance for machine learning projects, by filling the knowledge gap in the literature.This study has focused on applied machine learning projects in the Gothenburg area of Sweden, where data from 13 interview subjects from 10 corporations and organizations have been gathered.Cross-Industry Process for Data Mining (CRISP-DM), Team Data Science Process (TDSP), Scrum and Kanban have been found to be used in machine learning projects, but with modifications.Digital transformation and organizational change management have strong relevance for machine learning projects.The study found and elaborated on the three ML solution procurement options: in-house development, outsourcing and Commercial Off-the-Shelf (COTS) solutions.Finally, trust in the technology, the team and external collaborations were found important for machine learning project investments.