Managers putting Turing's ideas to the test: A critical success factor study about machine learning projects in software organizations

Sammanfattning: Through a qualitative multiple-case study in a comparative design, this thesis aims to discover the critical success factors for implementing machine learning (ML) for software companies. The empirical material consists of interviews and documents from four case organizations of both failed and successful projects. A theoretical framework based on the school of project management called critical success factor research is used to analyze the findings. The findings illustrate six critical success factors for ML implementations in software companies. Clear objectives and goals, Effective project management methodologies, and Realistic schedule have been found as success factors in both the software implementation literature and this study. Three new factors, Experimentation over planning, Deep understanding of the dataset, and Solution over technology orientation, have been found in this study. These differences indicate what project managers need to master in order to implement ML models in software organizations successfully. The study also increased the understanding of project success factors into a new context of the developing subfield of ML implementation.