rbcluster 0.0.1 → 0.0.2

data/README.md

@@ -3,6 +3,9 @@ rbcluster
 
 Ruby bindings to the Cluster C library.
 
+[![Build Status](https://secure.travis-ci.org/jarib/rbcluster.png)](http://travis-ci.org/jarib/rbcluster)
+
+
 TODO
 ----
 
@@ -10,13 +13,12 @@ Functions:
 
   * Cluster.clustercentroids
   * Cluster.clustermedoids
-  * Cluster::Tree#{cut,slice,[],fetch}
+  * Cluster::Tree#slice
 
 Other:
 
-* an examples/ folder
 * make :transpose work
-* specs for bad inputs
+* more specs for bad inputs
 
 DONE
 ----
@@ -36,6 +38,7 @@ See also
 
 * http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm
 * http://bonsai.hgc.jp/~mdehoon/software/cluster/cluster.pdf
+* http://github.com/jarib/rbcluster
 
 Note on Patches/Pull Requests
 -----------------------------

data/ext/rbcluster/rbcluster.c

@@ -10,6 +10,7 @@
 
 VALUE rbcluster_mCluster = Qnil;
 VALUE rbcluster_cNode = Qnil;
+VALUE rbcluster_cTree = Qnil;
 
 VALUE rbcluster_rows2rb(double** data, int nrows, int ncols) {
   VALUE rows = rb_ary_new2((long)nrows);
@@ -42,6 +43,26 @@ double* rbcluster_ary_to_doubles(VALUE data, int len) {
   return result;
 }
 
+Node* rbcluster_ary_to_nodes(VALUE data, int* len) {
+  Check_Type(data, T_ARRAY);
+
+  long length = RARRAY_LEN(data);
+
+  Node* result = (Node*)malloc(length*sizeof(Node));
+
+  for(int i = 0; i < length; ++i)
+  {
+    VALUE node = rb_ary_entry(data, i);
+
+    result[i].left = NUM2INT(rb_ivar_get(node, rb_intern("@left")));
+    result[i].right = NUM2INT(rb_ivar_get(node, rb_intern("@right")));
+    result[i].distance = NUM2DBL(rb_ivar_get(node, rb_intern("@distance")));
+  }
+
+  *len = (int)length;
+  return result;
+}
+
 double** rbcluster_ary_to_rows(VALUE data, int* nrows, int* ncols) {
   Check_Type(data, T_ARRAY);
   long rows, cols;
@@ -265,7 +286,7 @@ VALUE rbcluster_kcluster(int argc, VALUE* argv, VALUE self) {
   free(weight);
   free(clusterid);
 
-  return rb_ary_new3(3, result, rb_float_new(error), INT2NUM(ifound));
+  return rb_ary_new3(3, result, DBL2NUM(error), INT2NUM(ifound));
 }
 
 VALUE rbcluster_kmedoids(int argc, VALUE* argv, VALUE self) {
@@ -329,7 +350,7 @@ VALUE rbcluster_kmedoids(int argc, VALUE* argv, VALUE self) {
   free(clusterid);
   for(i = 1; i < nitems; ++i) free(distances[i]);
 
-  return rb_ary_new3(3, result, rb_float_new(error), INT2NUM(ifound));
+  return rb_ary_new3(3, result, DBL2NUM(error), INT2NUM(ifound));
 }
 
 VALUE rbcluster_median(VALUE self, VALUE ary) {
@@ -345,7 +366,7 @@ VALUE rbcluster_median(VALUE self, VALUE ary) {
     arr[i] = NUM2DBL(num);
   }
 
-  return rb_float_new(median((int)len, arr));
+  return DBL2NUM(median((int)len, arr));
 }
 
 VALUE rbcluster_mean(VALUE self, VALUE ary) {
@@ -361,7 +382,7 @@ VALUE rbcluster_mean(VALUE self, VALUE ary) {
     arr[i] = NUM2DBL(num);
   }
 
-  return rb_float_new(mean((int)len, arr));
+  return DBL2NUM(mean((int)len, arr));
 }
 
 VALUE rbcluster_distancematrix(int argc, VALUE* argv, VALUE self) {
@@ -403,7 +424,7 @@ VALUE rbcluster_distancematrix(int argc, VALUE* argv, VALUE self) {
       VALUE row = rb_ary_new();
 
       for(j = 0; j < i; ++j){
-        rb_ary_push(row, rb_float_new(distances[i][j]));
+        rb_ary_push(row, DBL2NUM(distances[i][j]));
       }
 
       // first row is NULL
@@ -482,7 +503,7 @@ VALUE rbcluster_clusterdistance(int argc, VALUE* argv, VALUE self) {
   rbcluster_free_rows(rows, nrows);
   rbcluster_free_mask(mask, nrows);
 
-  return rb_float_new(result);
+  return DBL2NUM(result);
 }
 
 VALUE rbcluster_create_node(Node* node) {
@@ -500,6 +521,10 @@ VALUE rbcluster_node_initialize(int argc, VALUE* argv, VALUE self) {
 
   rb_scan_args(argc, argv, "21", &left, &right, &distance);
 
+  if(NIL_P(distance)) {
+    distance = DBL2NUM(0.0);
+  }
+
   rb_ivar_set(self, rb_intern("@left"), left);
   rb_ivar_set(self, rb_intern("@right"), right);
   rb_ivar_set(self, rb_intern("@distance"), distance);
@@ -586,7 +611,8 @@ VALUE rbcluster_treecluster(int argc, VALUE* argv, VALUE self) {
   rbcluster_free_rows(rows, nrows);
   rbcluster_free_mask(mask, nrows);
 
-  return result;
+  VALUE args[] = { result };
+  return rb_class_new_instance(1, args, rbcluster_cTree);
 }
 
 VALUE rbcluster_somcluster(int argc, VALUE* argv, VALUE self) {
@@ -648,7 +674,7 @@ VALUE rbcluster_somcluster(int argc, VALUE* argv, VALUE self) {
     for(j = 0; j < nygrid; ++j) {
       jarr = rb_ary_new2(ncols);
       for(k = 0; k < ncols; ++k) {
-        rb_ary_push(jarr, rb_float_new(celldata[i][j][k]));
+        rb_ary_push(jarr, DBL2NUM(celldata[i][j][k]));
       }
       rb_ary_push(iarr, jarr);
     }
@@ -672,7 +698,7 @@ VALUE rbcluster_somcluster(int argc, VALUE* argv, VALUE self) {
   return rb_ary_new3(2, rb_clusterid, rb_celldata);
 }
 
-void print_doubles(double* vals, int len) {
+void rbcluster_print_doubles(double* vals, int len) {
   puts("[");
   for(int i = 0; i < len; ++i) {
     printf("\t%d: %f\n", i, vals[i]);
@@ -680,7 +706,7 @@ void print_doubles(double* vals, int len) {
   puts("]");
 }
 
-void print_double_matrix(double** vals, int nrows, int ncols) {
+void rbcluster_print_double_matrix(double** vals, int nrows, int ncols) {
   puts("[");
   for(int i = 0; i < nrows; ++i) {
     printf("\t[ ");
@@ -724,7 +750,7 @@ VALUE rbcluster_pca(VALUE self, VALUE data) {
 
   int ok = pca(nrows, ncols, u, v, w);
   if(ok == -1) {
-    rb_raise(rb_eStandardError, "could not allocate memory");
+    rb_raise(rb_eNoMemError, "could not allocate memory");
   } else if(ok > 0) {
     rb_raise(rb_eStandardError, "svd failed to converge");
   }
@@ -751,9 +777,37 @@ VALUE rbcluster_pca(VALUE self, VALUE data) {
   return rb_ary_new3(4, mean, coordinates, pc, eigenvalues);
 }
 
+VALUE rbcluster_cuttree(VALUE self, VALUE nodes, VALUE clusters) {
+  int nelements, nclusters;
+
+  nclusters = NUM2INT(clusters);
+
+  Node* cnodes = rbcluster_ary_to_nodes(nodes, &nelements);
+  int n = nelements + 1;
+
+  if(nclusters < 1) {
+    rb_raise(rb_eArgError, "nclusters must be >= 1");
+  }
+
+  if(nclusters > n) {
+    rb_raise(rb_eArgError, "more clusters requested than items available");
+  }
+
+  int clusterid[n];
+  cuttree(n, cnodes, nclusters, clusterid);
+  free(cnodes);
+
+  if(clusterid[0] == -1) {
+    rb_raise(rb_eNoMemError, "could not allocate memory for cuttree()");
+  }
+
+  return rbcluster_ints2rb(clusterid, (long)n);
+}
+
 void Init_rbcluster() {
   rbcluster_mCluster = rb_define_module("Cluster");
   rbcluster_cNode = rb_define_class_under(rbcluster_mCluster, "Node", rb_cObject);
+  rbcluster_cTree = rb_define_class_under(rbcluster_mCluster, "Tree", rb_cObject);
 
   rb_define_attr(rbcluster_cNode, "left", 1, 1);
   rb_define_attr(rbcluster_cNode, "right", 1, 1);
@@ -770,6 +824,7 @@ void Init_rbcluster() {
   rb_define_singleton_method(rbcluster_mCluster, "treecluster", rbcluster_treecluster, -1);
   rb_define_singleton_method(rbcluster_mCluster, "somcluster", rbcluster_somcluster, -1);
   rb_define_singleton_method(rbcluster_mCluster, "pca", rbcluster_pca, 1);
+  rb_define_singleton_method(rbcluster_mCluster, "cuttree", rbcluster_cuttree, 2);
 
   rb_define_const(rbcluster_mCluster, "C_VERSION", rb_str_new2(CLUSTERVERSION));
 }

data/lib/rbcluster.rb

@@ -1,5 +1,6 @@
 require 'rbcluster/version'
 require 'rbcluster/rbcluster'
 require 'rbcluster/tree'
+require 'rbcluster/node'
 
 RbCluster = Cluster

data/lib/rbcluster/node.rb

@@ -0,0 +1,7 @@
+module Cluster
+  class Node
+    def to_s
+      "(#{@left}, #{@right}): #{@distance}"
+    end
+  end
+end

data/lib/rbcluster/tree.rb

@@ -1,20 +1,47 @@
 module Cluster
   class Tree
     def initialize(nodes)
-      raise NotImplementedError, "patches welcome :)"
-
-      nodes.each_with_index do |node, idx|
+      @nodes = Array(nodes)
+      @nodes.each_with_index do |node, idx|
         unless node.kind_of?(Node)
           raise ArgumentError, "expected #{Node.class}, got #{node.class} at index #{idx}"
         end
       end
 
-      @nodes = nodes
     end
 
     def size
       @nodes.size
     end
 
+    def to_a
+      @nodes.dup
+    end
+
+    def to_s
+      @nodes.map { |e| "#{e}\n" }.join
+    end
+
+    def [](idx)
+      @nodes[idx]
+    end
+
+    def fetch(idx, &blk)
+      @nodes.fetch(idx, &blk)
+    end
+
+    def scale
+      max = @nodes.map { |e| e.distance }.max
+      @nodes.each do |node|
+        node.distance = node.distance /= max
+      end
+
+      nil
+    end
+
+    def cut(nclusters)
+      Cluster.cuttree(@nodes, nclusters)
+    end
+
   end
 end

data/lib/rbcluster/version.rb

@@ -1,3 +1,3 @@
 module Cluster
-  VERSION = "0.0.1"
+  VERSION = "0.0.2"
 end

data/rbcluster.gemspec

@@ -8,7 +8,7 @@ Gem::Specification.new do |s|
   s.platform    = Gem::Platform::RUBY
   s.authors     = ["Jari Bakken", "Michiel Jan Laurens de Hoon"]
   s.email       = ["jari.bakken@gmail.com"]
-  s.homepage    = "http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm"
+  s.homepage    = "http://github.com/jarib/rbcluster"
   s.summary     = %q{Ruby bindings for the Cluster C library}
   s.description = %q{This gem provides a Ruby extension to the clustering routines in the C Clustering Library (which also backs e.g. Python's pycluster and Perl's Algorithm::Cluster).}
 

data/spec/node_spec.rb

@@ -6,6 +6,7 @@ module Cluster
       n = Node.new(2, 3)
       n.left.should == 2
       n.right.should == 3
+      n.distance.should == 0.0
     end
 
     it "takes an optional distance" do

data/spec/tree_spec.rb

@@ -0,0 +1,58 @@
+require 'spec_helper'
+
+module Cluster
+  describe Tree do
+    context "creating" do
+      it "should raise ArgumentError if not given an array of Nodes" do
+        lambda { Tree.new(1) }.should raise_error(ArgumentError)
+        lambda { Tree.new([Node.new(1,2), Node.new(2,3), nil]) }.should raise_error(ArgumentError)
+        lambda { Tree.new }.should raise_error(ArgumentError)
+      end
+
+      it "returns a Tree instance when given an array of nodes" do
+        Tree.new([Node.new(1, 2)]).should be_kind_of(Tree)
+      end
+    end
+
+    context "using" do
+      let :tree do
+        Cluster.treecluster([
+            [  1.1, 2.2, 3.3, 4.4, 5.5],
+            [  3.1, 3.2, 1.3, 2.4, 1.5],
+            [  4.1, 2.2, 0.3, 5.4, 0.5],
+            [ 12.1, 2.0, 0.0, 5.0, 0.0]
+        ])
+      end
+
+      it "fetches a copy of the node array" do
+        arr = tree.to_a
+        arr.should be_kind_of(Array)
+        arr.size.should == 3
+        arr.clear
+
+        tree.size.should == 3
+      end
+
+      it "has a string representation" do
+        tree.to_s.should include('(2, 1): 2.6')
+      end
+
+      it "can scale the tree" do
+        tree.scale
+        tree.to_a.each { |n| n.distance.should be_between(0, 1)  }
+      end
+
+      it "can cut the tree" do
+        tree.cut(3).should == [1, 2, 2, 0]
+      end
+
+      it "gets a node" do
+        tree[0].should == tree.to_a[0]
+      end
+
+      it "fetches a node" do
+        tree.fetch(0).should == tree.to_a[0]
+      end
+    end
+  end
+end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: rbcluster
 version: !ruby/object:Gem::Version
-  version: 0.0.1
+  version: 0.0.2
   prerelease: 
 platform: ruby
 authors:
@@ -14,7 +14,7 @@ date: 2012-01-03 00:00:00.000000000Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rake-compiler
-  requirement: &2154075120 !ruby/object:Gem::Requirement
+  requirement: &2157667380 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ! '>='
@@ -22,10 +22,10 @@ dependencies:
         version: '0'
   type: :development
   prerelease: false
-  version_requirements: *2154075120
+  version_requirements: *2157667380
 - !ruby/object:Gem::Dependency
   name: rspec
-  requirement: &2154074620 !ruby/object:Gem::Requirement
+  requirement: &2157666880 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -33,7 +33,7 @@ dependencies:
         version: 2.6.0
   type: :development
   prerelease: false
-  version_requirements: *2154074620
+  version_requirements: *2157666880
 description: This gem provides a Ruby extension to the clustering routines in the
   C Clustering Library (which also backs e.g. Python's pycluster and Perl's Algorithm::Cluster).
 email:
@@ -55,13 +55,13 @@ files:
 - ext/rbcluster/extconf.rb
 - ext/rbcluster/rbcluster.c
 - lib/rbcluster.rb
+- lib/rbcluster/node.rb
 - lib/rbcluster/tree.rb
 - lib/rbcluster/version.rb
 - rbcluster.gemspec
 - spec/clustercentroids_spec.rb
 - spec/clusterdistance_spec.rb
 - spec/clustermedoids_spec.rb
-- spec/cuttree_spec.rb
 - spec/kcluster_spec.rb
 - spec/kmedoids_spec.rb
 - spec/median_mean_spec.rb
@@ -69,8 +69,9 @@ files:
 - spec/pca_spec.rb
 - spec/somcluster_spec.rb
 - spec/spec_helper.rb
+- spec/tree_spec.rb
 - spec/treecluster_spec.rb
-homepage: http://bonsai.hgc.jp/~mdehoon/software/cluster/software.htm
+homepage: http://github.com/jarib/rbcluster
 licenses: []
 post_install_message: 
 rdoc_options: []
@@ -98,7 +99,6 @@ test_files:
 - spec/clustercentroids_spec.rb
 - spec/clusterdistance_spec.rb
 - spec/clustermedoids_spec.rb
-- spec/cuttree_spec.rb
 - spec/kcluster_spec.rb
 - spec/kmedoids_spec.rb
 - spec/median_mean_spec.rb
@@ -106,5 +106,6 @@ test_files:
 - spec/pca_spec.rb
 - spec/somcluster_spec.rb
 - spec/spec_helper.rb
+- spec/tree_spec.rb
 - spec/treecluster_spec.rb
 has_rdoc: 

data/spec/cuttree_spec.rb

@@ -1,6 +0,0 @@
-require 'spec_helper'
-
-describe "Cluster.cuttree" do
-  pending
-end
-