readme

来自「经典cure聚类算法,实现用c++语言.大家」· 代码 · 共 38 行

0. I have modified an existing hierarchical clustering code and followed
   the CURE paper as closely as possible.  Please let me know ([e-mail]) 
   if you find any bugs in the code.  

1. First part of the input file (ex1.dat) is shown below.  Each row corresponds
   to (x,y) coordinates of a data point.

40.678 63.9831
41.4301 64.8066
41.3468 63.8069
42.3465 63.7236
41.93 62.9738
40.7637 62.8905

2. You compile and run the code as follows:

	% cluster -k 2 -a 0.1 -r 10 ex1.dat

    Note that -k option is for the number of clusters, -a is for alpha 
    parameter of CURE, and -r is the number of representative points of
    the cluster.

    The above run will generate 2 clusters with alpha 0.1 and the number
    of representative points 10.

3. This will generate ex1.dat-partition file like following (note that this
   is not an output from the real run):

0
0
1
1
1
0

    The first 2 data points and the last data point belongs to cluster 0 and
    the remaining 3 data points belong to cluster 1.