경영 의사결정 지원을 위한 프로세스 마이닝에서의 근본 원인 분석 = Root Cause Analysis in Process Mining for Managerial Decision Support

다국어 초록 (Multilingual Abstract)

With the widespread adoption of integrated enterprise information systems and the digitalization of business activities, process mining—analyzing organizational processes from event logs—has attracted increasing attention across industries. Conventional process mining, however, largely relies on frequency and pattern analyses, which are effective for revealing process flows and where anomalies occur but less effective at explaining why they occur. In management, identifying factors that causally drive outcomes is essential for deciding what to change and how to achieve performance improvement. Despite this need, systematic, context-aware root cause analysis in process mining remains limited. Prior causal approaches often search for causes using user-defined hypothesis templates, which can narrow the candidate space and weaken interpretability under time constraints and limited contextual information. As a result, important true causes may be missed, or spurious causes may be reported when context is ignored. To address these limitations, we propose a comprehensive root cause analysis methodology for process mining. We introduce three criteria for candidate cause identification—precedence, causal eligibility, and probability raising—and propose AR-ACE (Association Rule–Automatic Candidate Extraction), an association-rule-based method that automatically extracts candidate causes from event logs. We then apply causal tests to the extracted candidates to verify whether they constitute causal antecedents of the outcome. Comparative experiments with prior studies show that the proposed method identifies context-aware causes that existing approaches fail to detect. By automatically deriving evidence-based candidates from event logs, the proposed methodology reduces dependence on user subjectivity and provides actionable causal insights to support decision making for problem solving and process improvement.

목차 (Table of Contents)

• I. 서론 ·················································· 1
• 1. 연구 배경 및 목적 ··················································· 1
• 2. 논문 구성 ····························································· 6
• II. 배경 연구 ············································ 7
• 1. 프로세스 마이닝 ····················································· 7
• 1) 프로세스 마이닝 기법의 기본 개념 ···································· 7
• 2) 프로세스 마이닝 기법의 세 가지 기본 유형 ·························· 8
• 3) 프로세스 마이닝의 주요 구성 요소 ··································· 11
• 4) 프로세스 마이닝의 도출 알고리즘 ····································· 12
• (1) 알파 마이닝 ···························································· 12
• (2) 휴리스틱 마이닝 ······················································ 14
• (3) 퍼지 마이닝 ···························································· 17
• (4) 유전자 알고리즘 마이닝 ·············································· 19
• 2. 프로세스 마이닝에서의 원인 분석 ······························· 21
• Ⅲ. 선행 연구 ··········································· 24
• 1. 프로세스 마이닝의 연구 현황 ···································· 24
• 2. 프로세스 마이닝에서의 원인 분석 연구 현황 ················ 26
• 3. 인과 기반 원인 분석 ··············································· 28
• 4. 기존 연구와의 차이점 ·············································· 29
• Ⅳ. 방법론 ·············································· 31
• 1. 프로세스 도출 및 분석 ············································ 35
• 2. 원인 후보군 도출 ··················································· 36
• 3. 인과성 검정 및 원인 제시 ········································ 44
• 1) 인과 강도 계산 ··························································· 44
• 2) 인과성 검정 ······························································· 47
• Ⅴ. 제안된 모델 검증 ·································· 49
• 1. 프로세스 도출 및 분석 ············································ 49
• 1) 이벤트 로그 수집 ······················································· 49
• 2) 전처리 ····································································· 51
• 3) 프로세스 도출 및 흐름 분석 ··········································· 52
• 2. 원인 후보군 도출 ··················································· 56
• 3. 인과성 검정 및 원인 제시 ········································ 60
• 1) 인과 강도 계산 ··························································· 60
• 2) 인과성 검정 ······························································· 61
• 3) 유사 연구와의 비교 ······················································ 63
• Ⅵ. 요약 및 결론 ······································· 70
• 1. 연구 요약 및 시사점 ·············································· 70
• 2. 연구의 한계점 및 향후 연구 ······································ 71
• 참고문헌 ················································ 73
• Abstract ················································ 79
• 부록 ····················································· 81