#Implement unoptimized models
#----------------------------
if False:
model = second_model()
num_epochs = 100
batch_size = 32
hm = model.fit(X_train,
y_train,
validation_data=(X_test, y_test),
epochs=num_epochs,
batch_size=batch_size)
gut.plot_KerasHistory_metrics(hm, 'nhanes_keras_model_metrics')
scores = model.evaluate(X, Y)
print("\n%s: %.2f%%" % (model.metrics_names[1], scores[1] * 100))
prob_train = model.predict(X_train)
predictions_train = [round(x[0]) for x in prob_train]
prob_test = model.predict(X_test)
predictions_test = [round(x[0]) for x in prob_test]
pred_train_df = pd.concat([
pd.DataFrame(prob_train, columns=['PROB']),
pd.DataFrame(predictions_train, columns=['Y_PRED_TRAIN'])
],
axis=1)
pred_test_df = pd.concat([
pd.DataFrame(prob_test, columns=['PROB']),
model.predict_classes(X_test)
#model.predict(X_test)
# Log Loss over time
plt.plot(History.history['loss'])
plt.plot(History.history['val_loss'])
plt.title('Model Loss')
plt.ylabel('Loss')
plt.xlabel('Epochs')
plt.legend(['train', 'test'])
plt.show()
# Model accuracy over time
plt.plot(History.history['acc'])
plt.plot(History.history['val_acc'])
plt.title('Model Accuracy')
plt.ylabel('Accuracy')
plt.xlabel('Epochs')
plt.legend(['train', 'test'])
plt.show()
print(model.metrics_names)
model.evaluate(X_train,y_train)
# averaged loss and accuracy on train
model.evaluate(X_test,y_test)
# averaged loss and accuracy on text
x_train, x_val, y_train, y_val = train_test_split(x_train,
y_train,
train_size=0.8,
random_state=55)
# print(x_train.max)
# print(x_train.min)
#2. 모델링
model = XGBClassifier(n_job=-1, use_label_encoder=False)
#3. 컴파일, 훈련
model.fit(x_train, y_train)
#4. 평가, 예측
loss = model.evaluate(x_test, y_test)
print('loss : ', loss)
y_pred = model.predict(x_test[:10])
# print(y_pred)
print(y_test[:10])
print(np.argmax(y_test[:10], axis=-1))
#DNN
#(784, )
# loss : [0.09116600453853607, 0.9779000282287598]
# [7 2 1 0 4 1 4 9 5 9]
#PCA 154
# loss : [0.13378241658210754, 0.9748571515083313]
# [9 4 5 3 8 8 8 1 6 4]
#Oversampling
x_train_res_train, y_train_res_train = sm.fit_sample(
x_train[train], y_train[train])
# model fiting with train set and model validatation with validation data set
history = model.fit(x_train_res_train,
y_train_res_train,
validation_data=(x_train[test], y_train[test]),
epochs=2000,
batch_size=300,
verbose=0)
historySet.append(history)
scores = model.evaluate(x_train[test], y_train[test], verbose=0)
#predict test dataset.
predicted = model.predict(x_val)
predicted = (predicted > 0.5)
#making confuction matrix
cm = confusion_matrix(y_val, predicted)
fpr, tpr, thresholds = metrics.roc_curve(y_val, predicted)
tprs.append(interp(mean_fpr, fpr, tpr))
roc_auc = auc(fpr, tpr)
print('ROC_AUC')
print(roc_auc)
print('TPR')
print(tpr[1])
print('FPR')
#X_test = np.hstack((X_test, X_test[:, :5]**5))
#X_test = np.hstack((X_test, X_test[:, :5]**6))
X_all = np.concatenate((X_train, X_test))
mu = sum(X_all) / X_all.shape[0]
sigma = np.std(X_all, axis=0)
return (X_train - mu) / sigma, Y_train, (X_test - mu) / sigma
def Test(model, X_test):
result = model.predict(X_test, batch_size=100)
index_name = [i for i in range(1, X_test.shape[0] + 1)]
df = pd.DataFrame(index_name, columns=['id'])
df['label'] = pd.DataFrame(result.astype(int))
df.to_csv('output.csv', index=False)
if __name__ == '__main__':
X_train, Y_train, X_test = Read_data()
model = XGBClassifier()
#model.compile(loss='binary_crossentropy', optimizer=Adagrad(lr=0.01), metrics=['accuracy'])
model.fit(X_train,
Y_train,
batch_size=100,
epochs=50,
validation_split=0.1,
shuffle=True)
score = model.evaluate(X_train, Y_train)
print('Total loss: ', score[0])
print('Accuracy: ', score[1])
Test(model, X_test)
