GENDER_CLASSIFIER_USING MULTIPLE MODELS.py

# IN THIS SCRIPT I USED 4 DIFFERENT CLASSIFIER MODELS ALONG WITH DECISION TREE CLASSIFIER
# TO TRAIN A KNOWN DATA SET OF 'HEIGHT', 'WEIGHT' AND 'SHOE SIZE' OF TWO GENDERS
# THEN TEST AND PREDICT WHETHER IF CERTAIN MEASUREMENTS BELONG TO A MALE OR A FEMALE
# THEN I USE ACCURACY METRICS TO DETERMINE WHICH OF THE 5 CLASSIFIER MODELS HAVE THE BEST ACCURACY SCORE
# ALL USING SCIKIT-LEARN PACKAGES

from sklearn import tree #-decision tree classifier
from sklearn.svm import SVC #-Support Vector Classification (SVC classifier)
from sklearn.linear_model import Perceptron #-perception classifier
from sklearn.neighbors import KNeighborsClassifier #-KNeighbors classifier
from sklearn.naive_bayes import GaussianNB
from sklearn.metrics import accuracy_score #-to determine accuracy
import numpy as np #-for array operations
from warnings import simplefilter #-IGNORING the annoying FutureWarning by SK-Learn
simplefilter(action='ignore', category=FutureWarning)

# Data set (X) and gender labels (Y)
X = [[185, 82, 45], [166, 62, 37], [175, 72, 41], [155, 53, 38], [165, 63, 41], [192, 92, 45], [177, 65, 33],
     [177, 70, 40], [159, 55, 37], [171, 75, 42], [181, 85, 43]]

Y = ['MALE', 'FEMALE', 'MALE', 'FEMALE', 'MALE', 'MALE', 'FEMALE', 'FEMALE', 'FEMALE', 'MALE', 'MALE']

# CALLING CLASSIFIERS
class_tree = tree.DecisionTreeClassifier()
class_SVC = SVC()
class_perc = Perceptron()
class_KNN = KNeighborsClassifier()
class_GSNB = GaussianNB()


# TRAINING THE MODELS USING X, Y DATA SET
class_tree = class_tree.fit(X, Y)
class_SVC = class_SVC.fit(X, Y)
class_perc = class_perc.fit(X, Y)
class_KNN = class_KNN.fit(X, Y)
class_GSNB = class_GSNB.fit(X, Y)

# TESTING PREDICTION USING SAME DATA SET
predict_tree = class_tree.predict(X)
predict_SVC = class_SVC.predict(X)
predict_perc = class_perc.predict(X)
predict_KNN = class_KNN.predict(X)
predict_GSNB = class_GSNB.predict(X)

# PREDICTING USING DIFFERENT DATA SET
print('ATTEMPT 1: for height: 250, weight: 70, shoe size: 45')
print('Decision tree guessed: ', class_tree.predict([[250, 70, 45]]))
print('SVC guessed: ', class_SVC.predict([[250, 70, 45]]))
print('Perceptron guessed: ', class_perc.predict([[250, 70, 45]]))
print('KNN guessed: ', class_KNN.predict([[250, 70, 45]]))
print('GaussianNB guessed: ', class_GSNB.predict([[250, 70, 45]]))

print('\nATTEMPT 2: for height: 175, weight: 72, shoe size: 41')
print('Decision tree guessed: ', class_tree.predict([[175, 72, 41]]))
print('SVC guessed: ', class_SVC.predict([[175, 72, 41]]))
print('Perceptron guessed: ', class_perc.predict([[175, 72, 41]]))
print('KNN guessed: ', class_KNN.predict([[175, 72, 41]]))
print('GaussianNB guessed: ', class_GSNB.predict([[175, 72, 41]]))

print('\nATTEMPT 3: for height: 166, weight: 62, shoe size: 37')
print('Decision tree guessed: ', class_tree.predict([[166, 62, 37]]))
print('SVC guessed: ', class_SVC.predict([[166, 62, 37]]))
print('Perceptron guessed: ', class_perc.predict([[166, 62, 37]]))
print('KNN guessed: ', class_KNN.predict([[166, 62, 37]]))
print('GaussianNB guessed: ', class_GSNB.predict([[166, 62, 37]]))

print('\nATTEMPT 4: for height: 177, weight: 165, shoe size: 33')
print('Decision tree guessed: ', class_tree.predict([[177, 65, 33]]))
print('SVC guessed: ', class_SVC.predict([[177, 65, 33]]))
print('Perceptron guessed: ', class_perc.predict([[177, 65, 33]]))
print('KNN guessed: ', class_KNN.predict([[177, 65, 33]]))
print('GaussianNB guessed: ', class_GSNB.predict([[177, 65, 33]]))

print('\nATTEMPT 5: for height: 171, weight: 75, shoe size: 42')
print('Decision tree guessed: ', class_tree.predict([[171, 75, 42]]))
print('SVC guessed: ', class_SVC.predict([[171, 75, 42]]))
print('Perceptron guessed: ', class_perc.predict([[171, 75, 42]]))
print('KNN guessed: ', class_KNN.predict([[171, 75, 42]]))
print('GaussianNB guessed: ', class_GSNB.predict([[171, 75, 42]]))

# TESTING ACCURACY OF THOSE PREDICTION
print("\nTesting 11 more data set")

accu_tree = accuracy_score(Y, predict_tree) * 100 # times 100 to make percentage
print('\nDecision Tree Accuracy: {}'.format(accu_tree))

accu_SVC = accuracy_score(Y, predict_SVC) * 100
print('SVC Accuracy: {}'.format(accu_SVC))

accu_perc = accuracy_score(Y, predict_perc) * 100
print('Perceptron Accuracy: {}'.format(accu_perc))

accu_KNN = accuracy_score(Y, predict_KNN) * 100
print('KNeighbors accuracy: {}'.format(accu_KNN))

accu_GSNB = accuracy_score(Y, predict_GSNB) * 100
print('GaussianNB accuracy: {}'.format(accu_GSNB))

# CHOOSING THE MOST ACCURATE MODEL
index = np.argmax([accu_tree, accu_SVC, accu_perc, accu_KNN, accu_GSNB])
clf = {0: 'Decision tree', 1: 'SVC', 2: 'Perceptron', 3: 'KNeighbors', 4: 'GaussianNB'}
print('Best gender classifier model is: {}'.format(clf[index]))