fcm | Fuzzy c-means clustering algorithm |
-d <data> | name of the input data frame |
-dout <dataout> | output data frame |
-cout <cldata> | cluster frame |
-eps <s-float> | stopping criterion |
-imax <int> | maximum number of iterations |
-nclu <nclust> | number of clusters/prototypes |
[-my <c-float>] | clustering criterion |
This command executes the fuzzy c-means algorithm in order to divide the data set, <data>, into clusters. The centroids of the clusters are stored into data frame <dataout> and the classification information to the classified data <cldata>. The number of clusters is defined with the flag -nclu. If no clustering criterion is given put inputdata to nearest cluster.
Example (ex5.16): This example demonstrates the basic use of fcm.
NDA> load boston.dat NDA> select flds -f boston.indus boston.dis boston.crim boston.age NDA> fcm -d flds -dout clusdat -cout clucld -eps 0.0 -fi 3 -my 0.7 -nclu 4 -imax 100 ...