K-means clustering
See also.
K-means clustering.
require 'kmeans/pair'
require 'kmeans/pearson'
require 'kmeans/cluster'
hash = {
"test01"=> {"hoge"=>0, "fuga"=>1, "piyo"=>0 },
"test02"=> {"hoge"=>2, "fuga"=>1, "piyo"=>3 },
"test03"=> {"hoge"=>3, "fuga"=>0, "piyo"=>1 },
"test04"=> {"hoge"=>0, "fuga"=>2, "piyo"=>0 },
"test05"=> {"hoge"=>4, "fuga"=>2, "piyo"=>3 },
"test06"=> {"hoge"=>3, "fuga"=>1, "piyo"=>1 }
}
kmeans = Kmeans::Cluster.new(hash, {
:centroids => 3,
:loop_max => 10
})
# Kmeans Clustering
kmeans.make_cluster
# The results differ for each run
p kmeans.cluster.values #=> [["test01", "test04"], ["test02"], ["test03", "test05", "test06"]]Hierarchical clustering.
require 'kmeans/pair'
require 'kmeans/pearson'
require 'kmeans/hcluster'
require 'kmeans/dendrogram'
hash = {
"test01"=> {"hoge"=>0, "fuga"=>1, "piyo"=>0 },
"test02"=> {"hoge"=>2, "fuga"=>1, "piyo"=>3 },
"test03"=> {"hoge"=>3, "fuga"=>0, "piyo"=>1 },
"test04"=> {"hoge"=>0, "fuga"=>2, "piyo"=>0 },
"test05"=> {"hoge"=>4, "fuga"=>2, "piyo"=>3 },
"test06"=> {"hoge"=>5, "fuga"=>8, "piyo"=>2 },
"test07"=> {"hoge"=>2, "fuga"=>6, "piyo"=>1 },
"test08"=> {"hoge"=>3, "fuga"=>4, "piyo"=>1 },
"test09"=> {"hoge"=>2, "fuga"=>2, "piyo"=>1 },
"test10"=> {"hoge"=>1, "fuga"=>2, "piyo"=>1 }
}
# Hierarchical clustering
hcluster = Kmeans::HCluster.new
keys = hash.keys
values = hash.values.map {|x| x.values }
clust = hcluster.hcluster(values)
p hcluster.printclust(clust, keys) #=> "-\n -\n -\n test07\n -\n test01\n test04\n -\n test08\n -\n test06\n -\n test09\n test10\n -\n test02\n -\n test03\n test05\n"
# Draw Dendrogram (Using with RMagick)
dendrogram = Kmeans::Dendrogram.new(
:imagefile => 'demo.png'
)
dendrogram.drawdendrogram(clust, keys) # Create demo.png on your directoryFor more info, see spec files.
See doc/ChangeLog.
See doc/AUTHORS.
774
See the file doc/LICENSE.