diff options
Diffstat (limited to 'kmeans')
| -rwxr-xr-x | kmeans | 18 |
1 files changed, 9 insertions, 9 deletions
@@ -57,7 +57,7 @@ def update assignment, data end def main - cfg = Trollop::options do + conf = Trollop::options do 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 @@ -65,26 +65,26 @@ def main 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 = read_data cfg[:input] - k = cfg[:k] + data = read_data conf[:input] + k = conf[:k] centroids = nil - if cfg[:init] == 1 + if conf[:init] == 1 centroids = rand_init(data, k) else centroids = rand_means_init(data, k) end STDERR.write "\n k #{k}\n" - STDERR.write " input #{cfg[:input]}\n" - STDERR.write "iterations #{cfg[:max_iterations]}\n" - STDERR.write "max no ch. #{cfg[:max_no_change]}\n" - STDERR.write " init #{cfg[:init]}\n\n" + STDERR.write " input #{conf[:input]}\n" + STDERR.write "iterations #{conf[:max_iterations]}\n" + STDERR.write "max no ch. #{conf[:max_no_change]}\n" + STDERR.write " init #{conf[:init]}\n\n" assignment = nil prev_stats = [] stats = [] no_change = 0 max_no_change = 5 STDERR.write "expected cluster sz=#{data.size/k.to_f}\n\n" - 0.upto(cfg[:max_iterations]) do |i| + 0.upto(conf[:max_iterations]) do |i| s = "iteration #{i}" STDERR.write "#{s}\n#{'-'*s.size}\n" assignment = assign centroids, data |