Class: GLM::Base

Inherits:
Object
  • Object
show all
Defined in:
lib/glm/base.rb

Direct Known Subclasses

Linear, Logit

Constant Summary collapse

@@initial_weight = 
1

Class Method Summary collapse

Instance Method Summary collapse

Constructor Details

#initialize(x, y, alpha = 0.05) ⇒ Base

Returns a new instance of Base.



3
4
5
6
7
8
 
# File 'lib/glm/base.rb', line 3

def initialize(x,y,alpha = 0.05)
  @x = x
  @y = y
  @@alpha = alpha
  @theta = GSL::Vector.alloc(Array.new(x.size2, @@initial_weight))
end

Class Method Details

.g(eta) ⇒ Object

Canonical reponse function, intended to be overriden



59
60
61
 
# File 'lib/glm/base.rb', line 59

def self.g(eta)
  raise 'Canonical reponse function g(eta) undefined'
end

Instance Method Details

#a ⇒ Object

Log partition function a(eta), intended to be overriden



11
12
13
 
# File 'lib/glm/base.rb', line 11

def a
  raise 'Log partition function a(eta) undefined'
end

#b ⇒ Object

intended to be overriden



16
17
18
 
# File 'lib/glm/base.rb', line 16

def b
  raise 'b undefined'
end

#est(x) ⇒ Object

Estimator

Arguments:

x: a feature vector in Array

Returns:

Estimation



36
37
38
 
# File 'lib/glm/base.rb', line 36

def est(x)
  format(x)
end

#eta(x) ⇒ Object

Natural parameter eta



47
48
49
50
 
# File 'lib/glm/base.rb', line 47

def eta(x)
  tmp = @theta * x.transpose
  return tmp
end

#format(x) ⇒ Object 



20
21
22
23
24
25
26
27
28
29
 
# File 'lib/glm/base.rb', line 20

def format(x)
  if x.is_a? GSL::Vector
    return output(x)
  elsif x.is_a? GSL::Matrix
    tmp = GSL::Vector.alloc x.size1
    (0...x.size1).each {|i|
      tmp[i]= output(x.row(i))}
    return tmp
  end
end

#gradient(x, y, v) ⇒ Object

Gradient on one sample



64
65
66
67
68
 
# File 'lib/glm/base.rb', line 64

def gradient(x,y,v)
  tmp = h(v)
  res = (y - tmp) * x
  return res
end

#h(x) ⇒ Object

Hypothesis function, outputs E(y|theta, x), mean of y given x parameterized by theta

Parameters:

x: a feature vector

Returns:

E(y|theta, x)



75
76
77
78
 
# File 'lib/glm/base.rb', line 75

def h(x)
  tmp = eta(x)
  return self.class.g(tmp)
end

#output(x) ⇒ Object

Output estimation from E(y|theta,x) Need overriding, except for plain linear regression



42
43
44
 
# File 'lib/glm/base.rb', line 42

def output(x)
  return h(x)
end

#single_update ⇒ Object

A step based on one sample in stochastic gradient descent



81
82
83
 
# File 'lib/glm/base.rb', line 81

def single_update()
  
end

#sto_update ⇒ Object

One complete loop of stochastic gradient descend



86
87
88
89
90
91
92
93
94
 
# File 'lib/glm/base.rb', line 86

def sto_update()
  (0...(@x.size1)).each do |i|
    (0...(@x.size2)).each do |j|
      updates = gradient(@x[i,j], @y[i], @x.row(i))
      @theta[j] = @theta[j] + @@alpha * updates
    end
  end
  pp @theta
end

#T ⇒ Object

Sufficient statistic T



54
55
56
 
# File 'lib/glm/base.rb', line 54

def T
  return @y
end

#theta ⇒ Object 



96
97
98
 
# File 'lib/glm/base.rb', line 96

def theta()
  return @theta
end