require 'daru'
require 'statsample-glm'
require 'open-uri'

content = open('http://www.ats.ucla.edu/stat/data/binary.csv')
File.write('binary.csv', content.read)

df = Daru::DataFrame.from_csv "binary.csv"
df.vectors = Daru::Index.new([:admit, :gpa, :gre, :rank])
df

glm = Statsample::GLM::compute df, :admit, :logistic, constant: 1
c = glm.coefficients :hash

Daru::Vector.new(c).exp # Calling `#exp` on Daru::Vector exponentiates each element of the Vector.

e = Math::E
new_data = Daru::DataFrame.new({
  gre: [df[:gre].mean]*4,
  gpa: [df[:gpa].mean]*4,
  rank: df[:rank].factors
  })

new_data[:rankp] = new_data.collect(:row) do |x|
  1 / (1 + e ** -(c[:constant]  + x[:gre] * c[:gre] + x[:gpa] * c[:gpa] + x[:rank] * c[:rank]))
end

new_data.sort! [:rank]

require 'distribution'

# Create a normally distributed Vector with mean 30 and standard deviation 2
rng = Distribution::Normal.rng(30,2)
body_weight = Daru::Vector.new(20.times.map { rng.call }.sort)

# Populate chances of survival, assume that people with less body weight on average
# are less likely to survive.
survive = Daru::Vector.new [0,0,0,0,0,1,0,1,0,0,1,1,0,1,1,1,0,1,1,1]

df = Daru::DataFrame.new({
  body_weight: body_weight,
  survive: survive
})

glm    = Statsample::GLM.compute df, :survive, :logistic, constant: 1
coeffs = glm.coefficients :hash

e = Math::E
df[:survive_pred] = df[:body_weight].map { |x| 1 / (1 + e ** -(coeffs[:constant] + x*coeffs[:body_weight])) }
df

df.plot type: [:scatter,:line], x: [:body_weight]*2, y: [:survive_pred]*2 do |plot, diagram|
  plot.x_label "Body Weight"
  plot.y_label "Probability of Survival"
end