Keras Denses VS SKLearn MLP VS SKLearn Decision Tree

Machine_Learning_Comparison.py

# Arda Mavi

import keras
import numpy as np
from sklearn import tree
from random import shuffle
from keras.layers import Dense
from keras.models import Sequential
from sklearn.neural_network import MLPClassifier

# Getting data:

# Y is always the third of X.
# Train data example X: [0,2,3,1] & Y: 3
X, Y = [], []
for a in range(4):
    for b in range(4):
        for c in range(4):
            for d in range(4):
                X.append([a,b,c,d])
shuffle(X)

# Train-test split.
test = X[int(8*len(X)/10)+1:]
X = X[:int(8*len(X)/10)]
print('Data len:', len(X)+len(test))
for i in X:
    Y.append(i[2])
print('Data Ready')


# Evaluate Rate:
def evaluate():
    sonuc = model.predict(test)
    count = 0
    for i in range(len(test)):
        if test[i][2] == int(sonuc[i]):
            count += 1
    evaluate_rate = (count*100/len(test))
    return(evaluate_rate)


# Keras Sequential:
model = Sequential()
model.add(Dense(1, input_shape=(4,), activation='relu'))
model.compile(optimizer=keras.optimizers.rmsprop(lr=0.1, decay=1e-6), loss='mse')
model.fit(X, Y, epochs=80)
print('Keras Dense Accuracy: ',evaluate())

# SKLearn MLP:
model = MLPClassifier(solver='lbfgs', alpha=1e-5, hidden_layer_sizes=(1), random_state=1)
model.fit(X, Y)
print('SKLearn MLP Accuracy: ',evaluate())

# SKLearn Decision Tree
model = tree.DecisionTreeClassifier()
model.fit(X, Y)
print('SKLearn Decision Tree Accuracy: ',evaluate())