diff options
Diffstat (limited to 'kmeans')
| -rwxr-xr-x | kmeans | 17 |
1 files changed, 8 insertions, 9 deletions
@@ -1,12 +1,12 @@ #!/usr/bin/env ruby -require 'zipf' -require 'optimist' +require "zipf" +require "optimist" def read_data fn data = {} ReadFile.new(fn).readlines_strip.map{ |i| - a = i.split ' ', 2 + a = i.split " ", 2 v = SparseVector.from_kv a.last data[a.first] = v } @@ -30,7 +30,7 @@ end def assign centroids, data assignment = {} data.each_pair { |name,feature_vector| - min = 1.0/0 + min = Float::INFINITY min_index = nil centroids.each_with_index { |c,i| dist = c.euclidian_dist(feature_vector) @@ -61,10 +61,10 @@ def main opt :k, "k", :type => :int, :required => true opt :input, "input: one feature vector per line", :type => :string, :required => true opt :max_iterations, "max. number of iterations", :type => :int, :default => 100 - opt :max_no_change, "max. No of stalled iterations before stopping ", :type => :int, :short => '-n', :default => 3 - opt :init, "centroid initialization (1: sample k features vectors, 2: k-times do sample k feature and build mean)", :type => :int, :short => '-j', :default => 2 + opt :max_no_change, "max. No of stalled iterations before stopping ", :type => :int, :short => "-n", :default => 3 + opt :init, "centroid initialization (1: sample k features vectors, 2: k-times do sample k feature and build mean)", :type => :int, :short => "-j", :default => 2 end - # data is 'ID f1=v1 f2=v2' + # data is "ID f1=v1 f2=v2" data = read_data conf[:input] k = conf[:k] centroids = nil @@ -86,7 +86,7 @@ def main STDERR.write "expected cluster sz=#{data.size/k.to_f}\n\n" 0.upto(conf[:max_iterations]) do |i| s = "iteration #{i}" - STDERR.write "#{s}\n#{'-'*s.size}\n" + STDERR.write "#{s}\n#{"-" * s.size}\n" assignment = assign centroids, data sizes = [] assignment.each_pair { |centroid_index, a| @@ -114,4 +114,3 @@ def main end main - |