Publications
Detailed Information
고객 추천 시스템(CRM)에서 선험적 알고리즘과 LASSO의 비교 : A Study on comparative of Apriori Algorithm and LASSO in Customer Relationship Management(CRM)
| DC Field | Value | Language |
|---|---|---|
| dc.contributor.advisor | Myunghee Cho Paik | - |
| dc.contributor.author | 김종대 | - |
| dc.date.accessioned | 2017-07-19T08:44:58Z | - |
| dc.date.available | 2017-07-19T08:44:58Z | - |
| dc.date.issued | 2014-08 | - |
| dc.identifier.other | 000000021587 | - |
| dc.identifier.uri | https://hdl.handle.net/10371/131287 | - |
| dc.description | 학위논문 (석사)-- 서울대학교 대학원 : 통계학과, 2014. 8. Myunghee Cho Paik. | - |
| dc.description.abstract | 유통분야, 마케팅 또는 웹 마이닝 분야에서 상품 추천이나 고객들의 구매패턴의 연관성을 발견하기 위하여 연관규칙분석(association rule analysis) 알고리즘을 사용한다. 연관규칙분석은 대용량 데이터베이스에서 변수들 간의 흥미로운 관계를 찾도록 고안된 방법으로 자료에 존재하는 항목(item)들 간의 if-then 형식의 연관규칙을 찾는 방법으로서 비지도학습의 일종이다.
본 논문에서는 고객 상품 추천 시에 기존의 연관성 분석보다 효과가 좋은 모델을 탐구해보고 기존의 알고리즘과 비교·분석한다. 고객의 구매여부는 Binary형태이므로 로지스틱 회귀 분석을 도입하고, 구체적으로 Zou와 Hastie(2005)에 의해 제안된 elastic net 모델을 비교·평가한다. Elastic net 모델은 능형회귀와 Lasso회귀의 절충으로서, 상관관계가 있는 변수들 중에서 하나의 변수만을 흔히 선택하는 Lasso의 단점을 보완하는 형태이다. Elastic net모델을 구성하는 능형회귀와 Lasso의 가중치인 와 의 조정과, 기존의 알고리즘보다 효과가 좋은 모델의 지속적인 탐구를 통하여, 향후 CRM의 여러 분야에서 분석 및 예측을 실시하는데 다양한 전략을 구상할 수 있도록 큰 도움이 될 수 있게 한다. | - |
| dc.description.tableofcontents | 1. 서론 ······················································································· 1
2. 데이터 ··················································································· 3 2.1 데이터 설명 ········································································· 3 2.2 데이터 정제 ········································································· 5 2.2.1 구매빈도 합이 17미만인 item제거 ·········································· 5 2.2.2 상품명 변경 ···································································· 8 2.3 데이터 분할 ········································································· 8 2.3.1 Training set – Test set 분할 ············································ 8 2.3.2 Training item – Test item 분할 ········································· 8 3. 방법론 ··················································································· 9 3.1 연관 규칙 분석 ····································································· 9 3.1.1 연관 규칙 ······································································ 9 3.1.2 연관 규칙 분석의 척도 ······················································ 9 3.1.3 연관 규칙 분석의 절차 ····················································· 10 3.2 로지스틱 회귀 분석 ······························································· 11 3.3 벌점화 방법 ········································································ 12 3.3.1 능형회귀 ······································································ 13 3.3.2 Lasso 회귀 ·································································· 14 3.3.3 Elastic net 모델 ····························································· 15 4. 분석 및 결과 ·········································································· 16 4.1 Apriori 알고리즘 ································································· 16 4.1.1 Item 분할 전 ······························································· 16 4.1.2 Item 분할 후 ································································ 18 4.2 Elastic net 모델 ·································································· 18 4.2.1 Item 분할 전 ································································ 18 4.2.2 Item 분할 후 ································································ 20 4.3 결과 ·············································································· 21 5. 맺음말 ················································································ 23 | - |
| dc.format | application/pdf | - |
| dc.format.extent | 1917672 bytes | - |
| dc.format.medium | application/pdf | - |
| dc.language.iso | ko | - |
| dc.publisher | 서울대학교 대학원 | - |
| dc.subject | 빅데이터 | - |
| dc.subject | CRM | - |
| dc.subject | 연관성 분석 | - |
| dc.subject | Elastic net | - |
| dc.subject | Lasso | - |
| dc.subject.ddc | 519 | - |
| dc.title | 고객 추천 시스템(CRM)에서 선험적 알고리즘과 LASSO의 비교 | - |
| dc.title.alternative | A Study on comparative of Apriori Algorithm and LASSO in Customer Relationship Management(CRM) | - |
| dc.type | Thesis | - |
| dc.contributor.AlternativeAuthor | Jong Dae Kim | - |
| dc.description.degree | Master | - |
| dc.citation.pages | vii, 33 | - |
| dc.contributor.affiliation | 자연과학대학 통계학과 | - |
| dc.date.awarded | 2014-08 | - |
- Appears in Collections:
- Files in This Item:
Item View & Download Count
Items in S-Space are protected by copyright, with all rights reserved, unless otherwise indicated.