mv

1 files changed, 71 insertions, 0 deletions

diff --git a/log-reg b/log-reg
new file mode 100755
index 0000000..82dc353
--- /dev/null
+++ b/log-reg
@@ -0,0 +1,71 @@
+#!/usr/bin/env ruby
+
+require 'zipf'
+require 'matrix'
+require 'trollop'
+
+def read_data fn
+  f = ReadFile.new fn
+  data = []
+  while line = f.gets
+    line.strip!
+    a = []
+    a << 1.0
+    tokenize(line).each { |i| a << i.to_f }
+    v = Vector.elements a
+    data << v
+  end
+  return data
+end
+
+def dot x, y
+  r = 0.0
+  x.each_with_index { |_,j|
+    r += x[j] * y[j]
+  }
+  return r
+end
+
+def approx_eql x, y, eps=10**-10
+  return false if !x||!y
+  return false if x.size!=y.size
+  x.each_with_index { |_,i|
+    return false if (x[i]-y[i]).abs>eps 
+  }
+  return true
+end
+
+def main
+  conf = Trollop::options do
+    opt :input,         "input data",         :type => :string, :required => true
+    opt :output,        "1/0 output data",    :type => :string, :required => true
+  end
+  data = read_data conf[:input]
+  dim = data[0].size
+  zeros = [0.0]*dim
+  t = ReadFile.readlines(conf[:output]).map{ |i| i.to_f }
+  model = Vector.elements zeros
+  prev_model = nil
+  gradient = Vector.elements zeros
+  hessian = Matrix.build(dim,dim) { |i,j| 0.0 }
+  i = 0 
+  while true
+    i += 1
+    data.each_with_index { |x,j|
+      m = 1.0/(1+Math.exp(-dot(model, x)))
+      gradient += (m-t[j]) * x
+      hup = Matrix.column_vector(x) * Matrix.row_vector(x)
+      hessian += m*(1.0-m) * hup
+    }
+    gradient /= data.size
+    hessian /= data.size
+    model -= hessian.inverse * gradient
+    break if approx_eql model, prev_model
+    prev_model = model
+  end
+  STDERR.write "ran for #{i} iterations\n"
+  puts model.to_s
+end
+
+main
+