ruby_linear_regression 0.0.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/lib/ruby_linear_regression.rb +94 -0
- metadata +47 -0
checksums.yaml
ADDED
@@ -0,0 +1,7 @@
|
|
1
|
+
---
|
2
|
+
SHA1:
|
3
|
+
metadata.gz: fd88f39c66fb2efe50b976fe3b1f04e06b179d18
|
4
|
+
data.tar.gz: 83842d112a3d92d239a08422e750f616923b2942
|
5
|
+
SHA512:
|
6
|
+
metadata.gz: a57cf9ee217bebd8ef47d417068c2abef876592d909fcf3359c05d11b12ebc7a4891cb583cd96193637c0cdbe6fe14278c92649377d860c8e32c33217a1d7a6e
|
7
|
+
data.tar.gz: 8f8aa53d404036c3ee657f94ad9d15f99f2ba9bf3da6b9854e10144fea7f87432e7ea2cf3bc9d8fe87a52c84121f242e472afc7a582407d6acfd0e088cd6fd0a
|
@@ -0,0 +1,94 @@
|
|
1
|
+
require 'matrix'
|
2
|
+
|
3
|
+
class RubyLinearRegression
|
4
|
+
|
5
|
+
attr_reader :x,:y,:theta,:mu,:sigma
|
6
|
+
|
7
|
+
def initialize
|
8
|
+
@mu = 0
|
9
|
+
@sigma = 1
|
10
|
+
end
|
11
|
+
|
12
|
+
def load_training_data x_data, y_data
|
13
|
+
|
14
|
+
# normalize the x_data
|
15
|
+
x_data = normalize_data( x_data )
|
16
|
+
|
17
|
+
# add 1 column to our data
|
18
|
+
x_data = x_data.map { |r| [1].concat(r) }
|
19
|
+
|
20
|
+
# build our x Matrix & y Vector
|
21
|
+
@x = Matrix.rows( x_data )
|
22
|
+
@y = Matrix.rows( y_data.collect { |e| [e] } )
|
23
|
+
|
24
|
+
@theta = Matrix[[0],[0]]
|
25
|
+
end
|
26
|
+
|
27
|
+
def compute_cost
|
28
|
+
# Compute the mean squared cost / error function
|
29
|
+
# First use matrix multiplication and vector subtracton to find errors
|
30
|
+
errors = (@x * @theta) - @y
|
31
|
+
|
32
|
+
# Then square all errors
|
33
|
+
errors = errors.map { |e| e * e }
|
34
|
+
|
35
|
+
# Find the mean of the square errors
|
36
|
+
mean_square_error = 0.5 * (errors.inject{ |sum, e| sum + e }.to_f / errors.row_size)
|
37
|
+
|
38
|
+
return mean_square_error
|
39
|
+
end
|
40
|
+
|
41
|
+
def train_normal_equation
|
42
|
+
# Calculate the optimal theta using the normal equation
|
43
|
+
# theta = ( X' * X )^1 * X' * y
|
44
|
+
@theta = (@x.transpose * @x).inverse * @x.transpose * @y
|
45
|
+
|
46
|
+
return @theta
|
47
|
+
end
|
48
|
+
|
49
|
+
def predict data
|
50
|
+
|
51
|
+
# normalize
|
52
|
+
data.each_index do |i|
|
53
|
+
data[i] = (data[i] - @mu[i]) / @sigma[i].to_f
|
54
|
+
end
|
55
|
+
|
56
|
+
# add 1 column to prediction data
|
57
|
+
data = [1].concat( data )
|
58
|
+
|
59
|
+
# perform prediction
|
60
|
+
prediction = (Matrix[data] * @theta)[0,0].to_f
|
61
|
+
|
62
|
+
return prediction
|
63
|
+
|
64
|
+
end
|
65
|
+
|
66
|
+
private
|
67
|
+
def normalize_data x_data
|
68
|
+
|
69
|
+
row_size = x_data.size
|
70
|
+
column_count = x_data[0].is_a?( Array) ? x_data[0].size : 1
|
71
|
+
|
72
|
+
x_norm = Array.new(row_size)
|
73
|
+
@mu = Array.new(column_count)
|
74
|
+
@sigma = Array.new(column_count)
|
75
|
+
|
76
|
+
0.upto(column_count - 1) do |column|
|
77
|
+
column_data = x_data.map{ |e| e[column] }
|
78
|
+
@mu[column] = column_data.inject{ |sum, e| sum + e } / row_size
|
79
|
+
@sigma[column] = (column_data.max - column_data.min)
|
80
|
+
end
|
81
|
+
|
82
|
+
0.upto(row_size-1) do |row|
|
83
|
+
row_data = x_data[row]
|
84
|
+
x_norm[row] = Array.new(column_count)
|
85
|
+
row_data.each_index do |i|
|
86
|
+
x_norm[row][i] = (row_data[i] - @mu[i]) / @sigma[i].to_f
|
87
|
+
end
|
88
|
+
end
|
89
|
+
|
90
|
+
return x_norm
|
91
|
+
|
92
|
+
end
|
93
|
+
|
94
|
+
end
|
metadata
ADDED
@@ -0,0 +1,47 @@
|
|
1
|
+
--- !ruby/object:Gem::Specification
|
2
|
+
name: ruby_linear_regression
|
3
|
+
version: !ruby/object:Gem::Version
|
4
|
+
version: 0.0.1
|
5
|
+
platform: ruby
|
6
|
+
authors:
|
7
|
+
- Soren Blond Daugaard
|
8
|
+
autorequire:
|
9
|
+
bindir: bin
|
10
|
+
cert_chain: []
|
11
|
+
date: 2017-10-06 00:00:00.000000000 Z
|
12
|
+
dependencies: []
|
13
|
+
description: "An implementation of a linear regression machine learning algorithm
|
14
|
+
implemented in Ruby.\n This algorithm uses Ruby's Matrix implementation
|
15
|
+
and the normal equation to train the data to the best fit.\n The
|
16
|
+
algorithm works with multiple independent variables to predict a dependent variable. "
|
17
|
+
email: sbd@ineptum.dk
|
18
|
+
executables: []
|
19
|
+
extensions: []
|
20
|
+
extra_rdoc_files: []
|
21
|
+
files:
|
22
|
+
- lib/ruby_linear_regression.rb
|
23
|
+
homepage: https://github.com/daugaard/linear-regression
|
24
|
+
licenses:
|
25
|
+
- MIT
|
26
|
+
metadata: {}
|
27
|
+
post_install_message:
|
28
|
+
rdoc_options: []
|
29
|
+
require_paths:
|
30
|
+
- lib
|
31
|
+
required_ruby_version: !ruby/object:Gem::Requirement
|
32
|
+
requirements:
|
33
|
+
- - ">="
|
34
|
+
- !ruby/object:Gem::Version
|
35
|
+
version: '0'
|
36
|
+
required_rubygems_version: !ruby/object:Gem::Requirement
|
37
|
+
requirements:
|
38
|
+
- - ">="
|
39
|
+
- !ruby/object:Gem::Version
|
40
|
+
version: '0'
|
41
|
+
requirements: []
|
42
|
+
rubyforge_project:
|
43
|
+
rubygems_version: 2.6.8
|
44
|
+
signing_key:
|
45
|
+
specification_version: 4
|
46
|
+
summary: Linear regression implemented in Ruby.
|
47
|
+
test_files: []
|