gmlvq.py

import numpy as np
import Data_Generata
# import matplotlib.pyplot as plt
from sklearn import neighbors
from sklearn.model_selection import train_test_split
import pylab
import math

training_list = []
validation_list = []

n = 5000
p = 0.5
k = 10

weights = 'uniform'

differences = []
men = []
women = []
gaps = []

t = range(0,100)

intended_gap = 0.5

# for i in t:
#
#     intended_gap = 0.01 * i
#
#     X, y, men_approved, women_approved = Data_Generata.CreditData(5, 50000, 10, 20000, 7, 2).generate_credit_data(n,p,intended_gap)
#     actual_gap = men_approved - women_approved
#     gaps.append(actual_gap)
#     men.append(men_approved)
#     women.append(women_approved)
#     differences.append(intended_gap - actual_gap)
#
# pylab.plot(t, differences, label='intended - actual')
# pylab.plot(t, men, label='men')
# pylab.plot(t, women, label='women')
# pylab.plot(t, gaps, label='gaps')
# pylab.legend(loc='upper left')
# pylab.show()

X, y, men_approved, women_approved = Data_Generata.CreditData(5, 50000, 10, 20000, 7, 2).generate_credit_data_debug(n,p,intended_gap)
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1/3, random_state=42)
X_train_true, X_valid, y_train_true, y_valid = train_test_split(X_train, y_train, test_size=1/2, random_state=42)

clf_training = neighbors.KNeighborsClassifier(k, weights=weights)
clf_training.fit(X_train_true, y_train_true)

training_score = clf_training.score(X_train_true, y_train_true)
validation_score = clf_training.score(X_valid, y_valid)

training_list.append(training_score)
validation_list.append(validation_score)

print("training score : {score} with k = {k}".format(k=k, score=training_score))
print("validation score : {score} with k = {k}".format(k=k, score=validation_score))

clf_test = neighbors.KNeighborsClassifier(k, weights=weights)
clf_test.fit(X_train_true, y_train_true)

best_k_test_score = clf_test.score(X_test, y_test)

print("test score : {score} with k = {k}".format(k=k, score=best_k_test_score))