EnsambleClasificadores.py

from sklearn.model_selection import train_test_split
from sklearn.naive_bayes import GaussianNB
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.discriminant_analysis import LinearDiscriminantAnalysis
from sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis
from sklearn.neural_network import MLPClassifier
from sklearn import neighbors
from sklearn import svm

'''
Script para probar la tasa de reconocimiento de cada uno de los modelos de clasificación
'''

"""
    Preguntar si se desea Probar la tasa de reconocimiento de datos de training, o de datos nuevos nunca antes vistos
"""
tasa = input("¿Desea estimar la tasa de reconocimiento a partir de los mismos datos de training? S = SI, N = NO: ")
nam = tuple(list(range(360)))
X = pd.read_csv("trn_tst/X.csv", names = nam)
Y_vector = pd.read_csv("trn_tst/Y.csv", names=[0])
y = np.ravel(Y_vector)

if tasa == "S":
    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=1)

else:
    X_train = X
    y_train = y
    X_test = pd.read_csv("datos_nuevos/X_test.csv", names = nam)
    y_test_prev = pd.read_csv("datos_nuevos/Y_test.csv", names = [0])
    y_test = np.ravel(y_test_prev)

print("TEST ACCURACY")

"""
Nearest Neighbors Classification
"""
n_neighbors = 7
neigh = neighbors.KNeighborsClassifier(n_neighbors, weights='uniform')

y_pred_neigh = neigh.fit(X_train, y_train).predict(X_test)
accuracy_neigh = ((y_test == y_pred_neigh).sum() / X_test.shape[0]) * 100
print('Nearest Neighbors : %.2f de un total de %d' % (accuracy_neigh, X_test.shape[0]))

"""
Gaussian Naive Bayes
"""
gnb = GaussianNB()
y_pred_gnb = gnb.fit(X_train, y_train).predict(X_test)
accuracy_gnb = ((y_test == y_pred_gnb).sum() / X_test.shape[0]) * 100
print('Gaussian Naive Bayes: %.2f de un total de %d' % (accuracy_gnb, X_test.shape[0]))

"""
Linear Discriminant Analysis
"""
lda = LinearDiscriminantAnalysis()
y_pred_lda = lda.fit(X_train, y_train).predict(X_test)
accuracy_lda = ((y_test == y_pred_lda).sum() / X_test.shape[0]) * 100
print('Linear Discriminant Analysis: %.2f de un total de %d' % (accuracy_lda, X_test.shape[0]))

"""
Quadratic Discriminant Analysis
"""
qda = QuadraticDiscriminantAnalysis()
y_pred_qda = qda.fit(X_train, y_train).predict(X_test)
accuracy_qda = ((y_test == y_pred_qda).sum() / X_test.shape[0]) * 100
print('Quadratic Discriminant Analysis: %.2f de un total de %d' % (accuracy_qda, X_test.shape[0]))

"""
Neural Network
"""
neuNet = clf = MLPClassifier(solver='sgd', alpha=0.0001, random_state=1)  ## alpha=1e-5 hidden_layer_sizes=(5, 2)
y_pred_neuNet = neuNet.fit(X_train, y_train).predict(X_test)
accuracy_neuNet= ((y_test == y_pred_neuNet).sum() / X_test.shape[0]) * 100
print('Neural Network: %.2f de un total de %d' % (accuracy_neuNet, X_test.shape[0]))

"""
Support Vector Machine
"""
supVM = svm.SVC(kernel='linear', C=1)
y_pred_supVM = supVM.fit(X_train, y_train).predict(X_test)
accuracy_supVM = ((y_test == y_pred_supVM).sum() / X_test.shape[0]) * 100
print('Support Vector Machine: %.2f de un total de %d' % (accuracy_supVM, X_test.shape[0]))