M.S. Candidate: Kerem Nazlıel
Program: Information Systems
Date: 29.12.2022 / 14:00
Place: A-108
Abstract: Developments in IT, Cloud, Analytics, and related fields have created an abundance of Data Science technologies for practitioners, developers, and organizations to use. This abundance and variety complicate the Data Science technology selection and management processes for the analytics teams. When teams select and use improper tools and technologies, they encounter problems and inefficiencies, also known as technical debt. As a remedy, this thesis proposes a systematic technology selection method considering the analytics technology selection literature and tests it on a case study. This method consists of a survey with open-ended questions to determine the requirements of a given Data Science Workflow, linkage grids to map technologies to these requirements, and multi-criteria-decision-making to rank the technologies according to practitioners’ needs and preferences. This method enables decision-makers to compare the technology alternatives and select the most suitable Data Science Technology Stack. While the existing studies in this domain consider the technology selection problem in isolation and investigate a subset of technologies, the proposed method encapsulates the end-to-end Data Science Process and the entire analytics technology landscape considering the key principles for developing industrially relevant strategic technology management toolkits.