Improving operational decision-making through decision mining - utilizing method engineering for the creation of a decision mining method

Abstract

Context

This study addresses the challenge of enhancing the efficiency and agility of decision support software supporting both operational decision-making and software production teams developing decision support software. It centers on creating a method that assists in mining decisions, checking decisions on conformance, and improving decisions, which supports software production teams in developing decision support software.

Objective

The primary objective is to develop an explicit, clear, and structured approach for discovering, checking, and improving decisions using decision support software. The study aims to create a blueprint for software production teams to develop Decision Mining (DM) software, in line with recent advancements in the field. Additionally, it seeks to provide a consolidated, methodical overview of activities and deliverables in the DM research field.

Method

The research employs method engineering principles to construct a method for DM that leverages the existing body of knowledge by utilizing a Systematic Literature Review (SLR). The study focuses on developing individual building blocks and method fragments incorporated into seven DM scenarios.

Results

The study led to the creation of a Decision Mining Method (DMM), which includes 138 method fragments grouped into eleven categories. These fragments were systematically merged to form a comprehensive DMM. The method encapsulates the complexity of DM and provides practical applicability in real-world scenarios, highlighted by the identification of seven distinct scenarios in DM phases. The study also conducted the first SLR in the DM field, providing a comprehensive overview of current practices and outcomes.

Conclusion

The study helps in advancing the DM field by creating a structured approach and a comprehensive method for DM, aligning with recent developments in the field. It successfully aggregated the fragmented DM domain into a cohesive methodological overview, crucial for future research. The study also lays out a detailed agenda for future research, focusing on expanding and validating the DMM, incorporating cross-disciplinary insights, and addressing the challenges in machine learning within DM. The future research directions aim to refine and broaden the applicability of the DMM, ensuring its effectiveness in diverse practical contexts and contributing to a more holistic and comprehensive approach to decision mining.