linreg

3 files changed, 161 insertions, 0 deletions

diff --git a/linreg.rb b/linreg.rb
new file mode 100755
index 0000000..09630a6
--- /dev/null
+++ b/linreg.rb
@@ -0,0 +1,61 @@
+#!/usr/bin/env ruby
+
+require 'nlp_ruby'
+require 'trollop'
+
+
+def read_data fn
+  f = ReadFile.new fn
+  data = []
+  while line = f.gets
+    line.strip!
+    v = SparseVector.new
+    a = []
+    a << 1.0
+    tokenize(line).each { |i| a << i.to_f }
+    v.from_a(a)
+    data << v
+  end
+  return data  
+end
+
+def main
+  cfg = Trollop::options do
+    opt :input,         "input data",         :type => :string, :required => true
+    opt :output,        "output data",        :type => :string, :required => true
+    opt :learning_rate, "learning rate",      :type => :float,  :default => 0.07
+    opt :stop,          "stopping criterion", :type => :int,    :default => 100
+  end
+  data = read_data cfg[:input]
+  zeros = [0.0]*data[0].size
+  t = ReadFile.new(cfg[:output]).readlines.map{ |i| i.to_f }
+  model = SparseVector.new zeros
+  stop = 0
+  prev_model = nil
+  i = 0
+  while true
+    i += 1
+    u = SparseVector.new zeros
+    data.each_with_index { |d,j|
+      u += d * ((model.dot(d) - t[j])*(1.0/t.size))
+    }
+    u *= cfg[:learning_rate]
+    model -= u
+    if model.approx_eql? prev_model 
+      stop += 1
+    else
+      stop = 0
+    end
+    break if stop==cfg[:stop]
+    prev_model = model
+  end
+  tss = t.map{ |y| (y-t.mean)**2 }.sum
+  j = -1
+  rss = t.map{ |y| j+=1; (y-model.dot(data[j]))**2 }.sum
+  STDERR.write "ran for #{i} iterations\n R^2=#{1-(rss/tss)}\n"
+  puts model.to_s
+end
+
+
+main
+
diff --git a/test/linreg/x.dat b/test/linreg/x.dat
new file mode 100644
index 0000000..3d93394
--- /dev/null
+++ b/test/linreg/x.dat
@@ -0,0 +1,50 @@
+   2.0658746e+00
+   2.3684087e+00
+   2.5399929e+00
+   2.5420804e+00
+   2.5490790e+00
+   2.7866882e+00
+   2.9116825e+00
+   3.0356270e+00
+   3.1146696e+00
+   3.1582389e+00
+   3.3275944e+00
+   3.3793165e+00
+   3.4122006e+00
+   3.4215823e+00
+   3.5315732e+00
+   3.6393002e+00
+   3.6732537e+00
+   3.9256462e+00
+   4.0498646e+00
+   4.2483348e+00
+   4.3440052e+00
+   4.3826531e+00
+   4.4230602e+00
+   4.6102443e+00
+   4.6881183e+00
+   4.9777333e+00
+   5.0359967e+00
+   5.0684536e+00
+   5.4161491e+00
+   5.4395623e+00
+   5.4563207e+00
+   5.5698458e+00
+   5.6015729e+00
+   5.6877617e+00
+   5.7215602e+00
+   5.8538914e+00
+   6.1978026e+00
+   6.3510941e+00
+   6.4797033e+00
+   6.7383791e+00
+   6.8637686e+00
+   7.0223387e+00
+   7.0782373e+00
+   7.1514232e+00
+   7.4664023e+00
+   7.5973874e+00
+   7.7440717e+00
+   7.7729662e+00
+   7.8264514e+00
+   7.9306356e+00
diff --git a/test/linreg/y.dat b/test/linreg/y.dat
new file mode 100644
index 0000000..1f4f963
--- /dev/null
+++ b/test/linreg/y.dat
@@ -0,0 +1,50 @@
+   7.7918926e-01
+   9.1596757e-01
+   9.0538354e-01
+   9.0566138e-01
+   9.3898890e-01
+   9.6684740e-01
+   9.6436824e-01
+   9.1445939e-01
+   9.3933944e-01
+   9.6074971e-01
+   8.9837094e-01
+   9.1209739e-01
+   9.4238499e-01
+   9.6624578e-01
+   1.0526500e+00
+   1.0143791e+00
+   9.5969426e-01
+   9.6853716e-01
+   1.0766065e+00
+   1.1454978e+00
+   1.0340625e+00
+   1.0070009e+00
+   9.6683648e-01
+   1.0895919e+00
+   1.0634462e+00
+   1.1237239e+00
+   1.0323374e+00
+   1.0874452e+00
+   1.0702988e+00
+   1.1606493e+00
+   1.0778037e+00
+   1.1069758e+00
+   1.0971875e+00
+   1.1648603e+00
+   1.1411796e+00
+   1.0844156e+00
+   1.1252493e+00
+   1.1168341e+00
+   1.1970789e+00
+   1.2069462e+00
+   1.1251046e+00
+   1.1235672e+00
+   1.2132829e+00
+   1.2522652e+00
+   1.2497065e+00
+   1.1799706e+00
+   1.1897299e+00
+   1.3029934e+00
+   1.2601134e+00
+   1.2562267e+00