A data analytics methodology for a fast-changing discipline

  • E (Erik) Ingen van

Student thesis: Master's Thesis

Abstract

Massive investments are being made in Data Analytics (DA) but there is no substantial effort to mature the working methods. Around 82% of DA practitioners do not use any methodology in their projects, while 85% believe it would improve their work. Organizations that rely on ad hoc processes (as opposed to planned processes) are only half as likely to rate their projects as successful. This study helps DA practitioners to overcome this contradiction.
A design science research method was chosen in combination with case study research to design a new matrix of project and methodology classes and their connections. The matrix is strongly rooted in existing DA methodologies like CRISP-DM and Snail Shell with the difference that the matrix is layered. Fourteen DA experts were consulted in interviews and focus groups to reflect on a new matrix.
The research resulted in the design of project classes such as hypothesis generation/testing, big data, self-service analytics and data governance and related them to the process steps of the DA methodology. This helps DA practitioners to make an elaborated methodology choice while also being able to gear it to their preferred orientation in terms of agile, iterative or waterfall.
Date of Award28 Jun 2020
Original languageEnglish
SupervisorRob Kusters (Examinator) & Jeroen Baijens (Co-assessor)

Keywords

  • data analytics methodology
  • agile
  • self-service analytics
  • cloud native
  • pipeline
  • continuous integration

Cite this

'