星期二, 3月 06, 2018

CLUSTERING ANALYSIS

CLUSTERING ANALYSIS
  • Hierarchical
    • Distance
      • 歐式距離 (Euclidean distance): d(X,Y) = \sqrt{ \Sigma_i( x_i - y_i)^2 }
      • 曼哈頓距離 (Manhanttan distance): d(X,Y) =  \Sigma_i | ( x_i - y_i) |
      • 坎培拉距離 (Canberra distance): d(X,Y) = \Sigma_i \frac{  | ( x_i - y_i) |}{ |x_i| + |y_i|}
    • Classfied
      • 單一連結法 (Single Linkage): d_{A,B} = min_{ i \in A; j \in B(d_{ij})}  
      • 完全連結法 (Complete Linkage): d_{A,B} = max_{ i \in A; j \in B(d_{ij})}  
      • 平均法 (Average Linkage): d_{A,B} = \frac{1}{n_A n_B} \Sigma_{i \in A} \Sigma_{i \in B} d_{ij}
      • 中心法 (Centroid  Method): d_{A,B} = d( \bar{\bar{x}}_A , \bar{\bar{x}}_B) = \| \bar{x}_A - \bar{x}_B \| ^2
  • Nonhierarchical.
    • K-means
    • Self-organizing map

Reference:
  1. 小羊的研究筆記
  2. https://kknews.cc/tech/rrkly3x.html
  3. https://read01.com/zh-tw/LNRzxM.html#.Wp56Q-huaUk

沒有留言: