Created
March 3, 2013 05:51
-
-
Save jamiecook/5074741 to your computer and use it in GitHub Desktop.
test some linear regression shit yo!
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
require 'ml4r_test_case' | |
require 'ml4r/ml4r' | |
require 'ml4r/linear_regression' | |
class TestLinearRegression < Ml4rTestCase | |
DELTA = 0.0001 | |
def test_simple_regression | |
lr = Ml4r::OLSLinearRegression.new([[1.0],[2.0],[3.0]], [1.0, 2.0, 3.0]) | |
assert_kind_of(Ml4r::LinearRegression, lr) | |
parameters = lr.getParameterEstimates | |
assert_in_delta(1.0, parameters.first.first, DELTA, "Factors") | |
assert_in_delta(0.0, parameters.second, DELTA, "Constant") | |
assert_in_delta([1,2,3], lr.getPredictedYs.to_a, DELTA, "Predicted Ys") | |
assert_in_delta(1.0, lr.getRSquared, DELTA) | |
end | |
def test_fixed_constant | |
fixed_constant = 0.1 | |
lr = Ml4r::OLSLinearRegression.new([[1.0],[2.0],[3.0]], [1.0, 2.0, 3.0], fixed_constant) | |
params = lr.getParameterEstimates | |
assert_equal(fixed_constant, params.last) | |
assert_operator params.first.first, :<, 1.0 | |
end | |
def test_weighted_regression | |
lr = Ml4r::OLSLinearRegression.new([[4.0],[2.0],[3.0], [4.0]], [1.0, 2.0, 3.0, 4.0], [0.0, 1.0, 1.0, 1.0]) | |
est = lr.getParameterEstimates | |
assert_in_delta(1.0, est.first.first, DELTA, "Factors") | |
assert_in_delta(0.0, est.second, DELTA, "Constant") | |
end | |
def test_multi_dimensional_regression | |
lr = Ml4r::OLSLinearRegression.new([[1.2, 1.0],[2.0, 1.8],[3.0, 3.2], [4.0, 4.4]], [1.1, 1.9, 3.1, 4.2]) | |
params = lr.getParameterEstimates | |
assert_kind_of(Array, params) | |
assert_equal(2, params.size) | |
assert_kind_of(Array, params.first) | |
assert_kind_of(Float, params.last) | |
assert_in_delta([0.5, 0.5], params.first, DELTA) | |
end | |
def test_benchmark | |
require "linefit" | |
require 'benchmark' | |
xs = (1..100).map(&:to_f).to_a | |
ys = xs.map { |x| x + rand(3).to_f } | |
Benchmark.bm do |x| | |
n = 300 | |
line_fit_time = x.report('linefit') { | |
lineFit = LineFit.new | |
1.upto(n) { |i| | |
lineFit.setData(xs,ys) | |
intercept, slope = lineFit.coefficients | |
} | |
} | |
ml4r_time = x.report(' ML4R') { | |
1.upto(n) { | |
lr = Ml4r::OLSLinearRegression.new(xs,ys) | |
lr_slope, lr_intercept = lr.getParameterEstimates | |
} | |
} | |
# Should be faster than the linefit gem :) | |
assert_operator ml4r_time.utime, :<, line_fit_time.utime | |
end | |
end | |
end |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment