CanopyBuilder

Full name Microsoft.VisualBasic.DataMining.Clustering.CanopyBuilder Assembly Microsoft.VisualBasic.DataMining.Framework Members 7

initial for k-means

00 Remarks

与传统的聚类算法(比如K-means)不同，Canopy聚类最大的特点是不需要事先指定k值(即clustering的个数)，因此具有很大的实际应用价值。与其他聚类算法相比，Canopy聚类虽然精度较低，但其在速度上有很大优势，因此可以使用Canopy聚类先对数据进行“粗”聚类，得到k值，以及大致的K个初始质心，再使用K-means进行进一步“细”聚类。所以Canopy+K-means这种形式聚类算法聚类效果良好。

01 Syntax

Microsoft.VisualBasic.DataMining.Clustering.CanopyBuilder

02 Methods

Name	Overloads	Summary
.ctor	1
AverageDistance	2	得到平均距离
TotalDistance	1
SquareDist	1
KMeansSeeds	1
Solve	1

03 Members

method .ctor #

#ctor(IEnumerable(Of ClusterEntity), Double, Double)

Remarks

value of T1 should greater than T2, example as:

T1 = 8 and T2 = 4

Parameters

Name	Type	Description
`data`	`IEnumerable(Of ClusterEntity)`	-
`T1`	`Double`	-
`T2`	`Double`	-

method AverageDistance #

AverageDistance(ClusterEntity())

得到平均距离

Parameters

Name	Type	Description
`points`	ClusterEntity()	-

method AverageDistance overload 2 #

AverageDistance(Double, Double())

Parameters

Name	Type	Description
`pointSize`	`Double`	-
`parts`	`Double()`	sum(parts)

method TotalDistance #

TotalDistance(ClusterEntity, ClusterEntity())

method SquareDist #

SquareDist(Double(), Double())

method KMeansSeeds #

KMeansSeeds()

method Solve #

Solve()