class LogisticRegressionGradientDescent

# File lib/logisticregressiongradientdescent.rb, line 7
def fit(x,y,alpha,num_iter)
  theta_start = N.zeros([x.cols,1])
  theta_end = cost_function(x,y,theta_start)
  @@theta_class = theta_end
end

# File lib/logisticregressiongradientdescent.rb, line 15
def cost_function(x,y,theta)
  m = y.rows
  h = x.dot(theta)
  sig_p = sigmoid(h)
  cost = (-y).transpose.dot(sig_p.log) - (N.ones([y.rows,y.cols])-y).transpose.dot((N.ones([sig_p.rows,sig_p.cols])-sig_p).log)
  cost_sum = cost.sum[0,0]*(1.0/m)
  grad = ((sig_p-y).transpose.dot(x))*(1.0/m)
  theta = grad
  return cost_sum,theta
end

# File lib/logisticregressiongradientdescent.rb, line 26
def sigmoid(z)
  g = N.zeros([z.cols,1])
  g = N.ones([z.rows,z.cols])/(N.ones([z.rows,z.cols]) + ((-z).exp))
  return g
end