def modelo2(train, test, coluna):
X_train = train.drop('default payment next month', axis=1)
Y_train = train['default payment next month']
X_test = test.drop('default payment next month', axis=1)
Y_test = test['default payment next month']
tamanho = len(X_train.columns) + 1
X_train = X_train.values
Y_train = Y_train.values
X_test = X_test.values
Y_test = Y_test.values
salvar = pandas.read_csv('resultados_stepwise.csv')
i = len(salvar.index) + 1
layer_1 = 46
layer_2 = 12 #se zero, comentar a linha do layer_2
rate = 0.01
epoch = 200 #50
batch = 128
decay = 1e-12
#decay = 0.0
drop = 0.5 #se 0, comentar a linha do dropout
model = load_model('my_model.h5')
predictions = model.predict(X_test)
predictions = pandas.DataFrame(predictions, columns=['SCORE'])
s = pandas.DataFrame(Y_test, columns=['CLASSE'])
x = predictions.join(s)
ks_max = teste_ks.KS(x)
salvar.loc[i] = [
ks_max, layer_1, layer_2, rate, epoch, batch, decay, drop, coluna
]
salvar.to_csv("resultados_stepwise.csv", index=False)
return ks_max
def modelo(train, test, coluna):
###
X_train = train.drop('default payment next month', axis=1)
Y_train = train['default payment next month']
X_test = test.drop('default payment next month', axis=1)
Y_test = test['default payment next month']
tamanho = len(X_train.columns) + 1
X_train = X_train.values
Y_train = Y_train.values
X_test = X_test.values
Y_test = Y_test.values
##
sm = SMOTE(kind='regular')
X_resampled, Y_resampled = sm.fit_sample(X_train, Y_train)
layer_1 = 46
layer_2 = 12 #se zero, comentar a linha do layer_2
rate = 0.01
epoch = 200 #50
batch = 128
decay = 1e-12
#decay = 0.0
drop = 0.5 #se 0, comentar a linha do dropout
salvar = pandas.read_csv('resultados_stepwise.csv')
i = len(salvar.index) + 1
# Model Creation, 1 input layer, 1 hidden layer and 1 exit layter
model = Sequential()
model.add(Dense(layer_1, input_dim=tamanho - 1, init='uniform'))
model.add(Activation('relu'))
model.add(Dropout(drop))
model.add(Dense(layer_2, init='uniform'))
# activation='relu'
model.add(Activation('relu'))
model.add(Dropout(drop))
#model.add(Dense(60, init='uniform'))
#model.add(Activation('relu'))
model.add(Dense(1, init='uniform'))
model.add(Activation('sigmoid'))
#activation='relu'
opt = keras.optimizers.SGD(lr=rate, decay=decay)
# Compile model
model.compile(loss='binary_crossentropy',
optimizer=opt,
metrics=['accuracy'])
# creating .fit
model.fit(X_resampled,
Y_resampled,
nb_epoch=epoch,
batch_size=batch,
verbose=0)
scores = model.evaluate(X_test, Y_test)
print()
print("%s: %.2f%%" % (model.metrics_names[1], scores[1] * 100))
# predicting model
predictions = model.predict(X_test)
predictions = pandas.DataFrame(predictions, columns=['SCORE'])
s = pandas.DataFrame(Y_test, columns=['CLASSE'])
x = predictions.join(s)
ks_max = teste_ks.KS(x)
salvar.loc[i] = [
ks_max, layer_1, layer_2, rate, epoch, batch, decay, drop, coluna
]
salvar.to_csv("resultados_stepwise.csv", index=False)
return ks_max