3.1.1. Hierarchical-entropy value
The hierarchical-entropy value of a node vb can be denoted as Hentropy(vb), and can be computed with the following equation:
Note that p(i,j) is the probability that an arbitrary tuple in D(vb) belongs to concept label L(i,j), and p(i,j) can be estimated by p๐i;j ผ
jD๐b;i;j j
jD๐vbj.
Let wi be the weight of level Li in tree CHT. Since L1 is the root label of CHT, the weight of L1 is set to w1 = 0. Assume w1, w2,
w3,. . . ,wh is an arithmetic sequence and Phi ผ1wi ผ 1. The following relation can then be used to compute wi. Table 6 shows the value of wi when h is varied from 3 to 5.
3.1.1. Hierarchical-entropy valueThe hierarchical-entropy value of a node vb can be denoted as Hentropy(vb), and can be computed with the following equation: Note that p(i,j) is the probability that an arbitrary tuple in D(vb) belongs to concept label L(i,j), and p(i,j) can be estimated by p๐i;j ผjD๐b;i;j jjD๐vbj.Let wi be the weight of level Li in tree CHT. Since L1 is the root label of CHT, the weight of L1 is set to w1 = 0. Assume w1, w2,w3,. . . ,wh is an arithmetic sequence and Phi ผ1wi ผ 1. The following relation can then be used to compute wi. Table 6 shows the value of wi when h is varied from 3 to 5.
การแปล กรุณารอสักครู่..