Notice
Recent Posts
Recent Comments
Link
«   2024/04   »
1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30
Archives
Today
Total
04-18 18:36
관리 메뉴

zyint's blog

Jaccard similarity coefficient 본문

예전글들

Jaccard similarity coefficient

진트­ 2009. 9. 21. 16:16

Jaccard Index(Jaccard similarity coefficient)는 두 집합 간의 유사도를 구하는 방식입니다.

 

\[J(A,B) = \frac{{\left| {A \cap B} \right|}} {{\left| {A \cup B} \right|}}\]

 

 

asymmetric binary attribute

두 개의 object A, B가 주어져 있고, 각각 n자리의 binary attributes일 때,
Jaccard coefficient는 두 object의 유사도를 구하는데 유용하게 사용됩니다.

  • \[{M_{11}}\] : A, B 모두 1인 자리수의 개수
  • \[{M_{01}}\] : A는 0이고 B가 1인 자리수의 개수
  • \[{M_{10}}\]: A는 1이고 B가 0인 자리수의 개수
  • \[{M_{00}}\]: A는 0이고 B가 0인 자리수의 개수
  • \[n = {M_{11}} + {M_{01}} + {M_{10}} + {M_{00}}\]

 

\[J = \frac{{{M_{11}}}} {{{M_{01}} + {M_{10}} + {M_{11}}}}\]

 

 

같이보기

Jaccard distance

 

 

 

참고자료

Jaccard index, Wikipedia

 

이 글은 스프링노트에서 작성되었습니다.

Comments