A carregar...
Publicação
Designing a collaborative AutoML tool to help organizations become data-driven
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 1.11 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
This study aims to address a lack of knowledge in the emerging field of automated machine learning
(AutoML) techniques. While the AutoML technology develops further and further and provides
increasingly robust and interesting results, there is only little to no current research on how this
technology can be adopted and scaled across different functions and teams of any organization.
Thus, this study raises the research question of how an information system that leverages AutoML
techniques can empower organizations and their non-technical individuals to collaborate on and
adopt machine learning techniques in their daily lives to unlock the value of available data. To gain a
clear analytical lens, this study is conducted in the environment of Management Consulting
Companies (MCCs) as they span all industries and multiple tasks within diverse organizations and
therefore promise a good transfer of knowledge to other application areas. A special emphasis is
given to non-technical users and the possibilities of them participating in such a system as that has
the potential to reach a large number of real-world practitioners. The identified problem is tackled
with a Design Science Research (DSR) approach. A workflow of how an information system can
support its users to leverage AutoML serves as an artifact that is evaluated by experts. Learnings
from the theory behind the proposal and its evaluation contribute to literature around AutoML and
the transformation of the MCC industry as well as practical applications in both fields. Results suggest
that AutoML is best used to conduct quick experiments and find out which applications have the
highest business value before involving experts. Major challenges are to help non-technical users
define a use case and prepare data.
Descrição
Dissertation presented as the partial requirement for obtaining a Master's degree in Information Management, specialization in Information Systems and Technologies Management
Palavras-chave
AutoML collaboration Data-driven decision making Machine Learning Design Science Research