Added Scaling

1 files changed, 64 insertions, 17 deletions

diff --git a/transform.rb b/transform.rb
index 73b803c..cfedd2b 100644
--- a/transform.rb
+++ b/transform.rb
@@ -76,38 +76,85 @@ class TransformTest < Test::Unit::TestCase
 
   def test_svd
 
-    m = GSL::Matrix[
-            [5,5,0,5],
-            [5,0,3,4],
-            [3,4,0,3],
-            [0,0,5,3],
-            [5,4,4,5],
-            [5,4,5,5] 
+     m = GSL::Matrix[
+       [5,5,0,5],
+       [5,0,3,4],
+       [3,4,0,3],
+       [0,0,5,3],
+       [5,4,4,5],
+       [5,4,5,5] 
      ]
+     foo = GSL::Matrix[[5,5,3,0,5,5]]
+     bar = GSL::Matrix[[5,4,5,5]]
+
+     #puts
+     #puts m.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+     #puts
+     #puts foo.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+     #puts
+     #puts bar.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+
+     # AutoScale to improve on representation
+     nr_cases, nr_features = m.size1, m.size2
+     (0..nr_features-1).each { |i|
+        autoscaler = OpenTox::Transform::AutoScale.new(m.col(i))
+        m.col(i)[0..nr_cases-1] = autoscaler.vs
+        bar.col(i)[0..0] = autoscaler.transform bar.col(i)
+     }
+     autoscaler = OpenTox::Transform::AutoScale.new(foo.transpose.col(0))
+     foo = GSL::Matrix[autoscaler.vs]
+
+     #puts
+     #puts m.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+     #puts
+     #puts foo.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+     #puts
+     #puts bar.to_a.collect { |r| r.collect{ |v| sprintf("%.2f", v) }.join(", ") }.join("\n")
+
+     # run SVD
      svd = OpenTox::Algorithm::Transform::SVD.new m
 
-     foo = svd.transform_feature GSL::Matrix[[5,5,3,0,5,5]]
+     foo = svd.transform_feature foo
      sim = []
      svd.vk.each_row { |x|
        sim << OpenTox::Algorithm::Similarity.cosine_num(x,foo.row(0))
      }
+
+     # # # NO AUTOSCALE
+     #assert_equal sim[0].round_to(3), 1.000
+     #assert_equal sim[1].round_to(3), 0.874
+     #assert_equal sim[2].round_to(3), 0.064
+     #assert_equal sim[3].round_to(3), 0.895
+
+     # # # AUTOSCALE
      assert_equal sim[0].round_to(3), 1.000
-     assert_equal sim[1].round_to(3), 0.874
-     assert_equal sim[2].round_to(3), 0.064
-     assert_equal sim[3].round_to(3), 0.895
+     assert_equal sim[1].round_to(3), 0.705
+     assert_equal sim[2].round_to(3), 0.023
+     assert_equal sim[3].round_to(3), 0.934
+
 
-     bar = svd.transform_instance GSL::Matrix[[5,4,5,5]]
+     bar = svd.transform_instance bar 
      sim = []
      svd.uk.each_row { |x|
        sim << OpenTox::Algorithm::Similarity.cosine_num(x,bar.row(0))
      }
 
-     assert_equal sim[0].round_to(3), 0.346
-     assert_equal sim[1].round_to(3), 0.966
-     assert_equal sim[2].round_to(3), 0.282
-     assert_equal sim[3].round_to(3), 0.599
-     assert_equal sim[4].round_to(3), 0.975
+     # # # NO AUTOSCALE
+     #assert_equal sim[0].round_to(3), 0.346
+     #assert_equal sim[1].round_to(3), 0.966
+     #assert_equal sim[2].round_to(3), 0.282
+     #assert_equal sim[3].round_to(3), 0.599
+     #assert_equal sim[4].round_to(3), 0.975
+     #assert_equal sim[5].round_to(3), 1.000 
+
+     # # # AUTOSCALE
+     assert_equal sim[0].round_to(3), -0.115
+     assert_equal sim[1].round_to(3), 0.425
+     assert_equal sim[2].round_to(3), -0.931
+     assert_equal sim[3].round_to(3), -0.352
+     assert_equal sim[4].round_to(3), 0.972
      assert_equal sim[5].round_to(3), 1.000 
+ 
   end
   
   def test_logas