Chapter 00 서장저술 목적··················································· 10운영시스템················································· 14개발 환경··················································· 14 Chapter 01 가설 검정서론·························································· 26가설과 가설 검정·········································· 26가설······················································· 26가설 검정················································ 27검증의 단계················································ 281단계: 귀무가설 및 대립가설 설명················· 282단계: 데이터 수집···································· 293단계: 통계 테스트 수행······························ 294단계: 귀무가설 기각 여부 결정···················· 305단계: 연구 결과 제시································ 30검증 오류················································ 31가설 검정 사례············································ 32데이터 로드············································· 32정규성 검증에 대한 가설 검정······················· 34상관성 검증에 대한 가설 검정······················· 36모수 통계 가설 검정··································· 39비모수 통계 가설 검정································ 44결론·························································· 48 Chapter 02 선형 회귀 모델링서론·························································· 50모델과 모델링············································· 50데이터셋···················································· 51단순 회귀 분석············································ 54가설설정················································· 54모델링···················································· 54모델링 결과············································· 54AIC······················································· 60다중 회귀 분석············································ 82모델링···················································· 88모델링 결과············································· 88회귀 모델 가정 검정··································· 92결론·························································104Chapter 03 이산 회귀 모델링서론·························································106모델링 기법···············································106로짓(Logit) 모형····································107프로빗 모형···········································107로짓과 프로빗 모형의 차이점······················108데이터 분석 사례·········································109Step 1: 라이브러리 가져오기·····················109Step 2: 데이터 로딩 및 이해······················109Step 3: 가설 설정···································110Step 4: 데이터 준비································110Step 5: Logit 모델링·······························113Step 6: Probit 모델링·····························116결론·························································118 Chapter 04 인과 추론 분석서론·························································120인과 추론의 4 단계······································121모델에서 목표 추정치 식별·························124확인된 추정치를 기반으로 인과 추론·············125획득한 추정치에 대한 반박·························126DoWhy 인과 추론의 특징·····························128명시적 식별 가능·····································128식별과 추정의 분리··································128자동화된 견고성 검사·······························128확장성··················································129인과 추론 분석 사례 - 호텔 예약 취소···············129Step 1: 라이브러리 가져오기·····················130Step 2: 데이터 로딩 및 데이터 이해·············130Step 3: 데이터 준비································133Step 4: DoWhy를 활용한 인과 관계 추정····142결론·························································151 Chapter 05 인과 발견 분석서론·························································154패키지 설치···············································154분석 방법 이해···········································155Step 1: 라이브러리 가져오기·····················156Step 2: 검증 데이터 생성··························156Step 3: 인과 관계 발견····························158Step 4: 오차 변수 간의 독립성 검증·············159분류 문제의 인과 발견··································160Step 1: 라이브러리 가져오기·····················160Step 2: 커스텀 함수 만들기·······················161Step 3: 데이터 로딩하기···························161Step 4: 모델링 하기································162Step 5: 변수 오차 간 독립성 검증···············165Step 6: 예측 모델 생성과 예측 영향도 분석···166수치 예측 문제의 인과 발견····························167Step 1: 라이브러리 가져오기·····················167Step 2: 커스텀 함수 만들기·······················168Step 3: 데이터 로딩하기···························168Step 4: 모델링 하기································170Step 5: 변수 오차 간 독립성 검증··············· 172Step 6: 예측 모델 생성과 예측 영향도 분석··· 172Step 7: 최적 개입의 추정·························· 173결론·························································174 Chapter 06 인과 영향 분석서론·························································176Causal Impact··········································177모델의 동작 방식의 이해···························· 179폭스바겐 인과 영향 분석 사례·························188Step 1: 라이브러리 로딩··························· 188Step 2: 데이터 로딩 및 데이터 이해············· 189Step 3: 기본 모델 분석···························· 191Step 4: 시계열 성분 분해·························· 195Step 5: 사용자 정의 모델·························· 197결론·························································202 Chapter 07 반대사실 분석서론·························································206소득 분류 반대사실 분석·······························207Step 1: 라이브러리 가져오기····················· 207Step 2: 데이터셋 로딩 및 이해··················· 207Step 3: DiCE로 카운터 팩트 생성··············· 209Step 4: 카운터 팩츄얼 사례 기반 속성 중요도··· 216주택 가격 예측 반대사실 분석 사례··················219Step 1: 라이브러리 로딩···························219Step 2: 데이터 로딩 및 이해······················220Step 3: DiCE로 카운터 팩트 생성···············223Step 4: 카운터 팩츄얼 기반 속성 중요도·······225결론·························································228참고문헌···················································229 색인·························································232