Обработка данных с использованием нейронной сети?

# first neural network with keras make predictions
from numpy import loadtxt
from keras.models import Sequential
from keras.layers import Dense
from keras.layers import LeakyReLU
 
# load the dataset
dataset = loadtxt('KQ.txt', delimiter=',')
 
# split into input (X) and output (y) variables
X = dataset[:,0:2]
y = dataset[:,2]
 
# define the keras model
model = Sequential()
model.add(Dense(6, input_dim=2, activation='softsign'))
model.add(Dense(1, LeakyReLU(alpha=0.1)))
 
# compile the keras model
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])
 
# fit the keras model on the dataset
model.fit(X, y, epochs=150, batch_size=10, verbose=0)
 
# evaluate the keras model
ll, accuracy = model.evaluate(X, y)
print('Loss: %.2f' % ll)
print('Accuracy: %.2f' % (accuracy*100))
 
# make class predictions with the model
predictions = model.predict(X)
 
# summarize the first 5 cases
for i in range(10):
    print('%s => %0.4f (expected %0.4f)' % (X[i].tolist(), predictions[i], y[i]))

mean = np.mean(X, axis = 0)
std = np.std(X, axis = 0)
X = [(x - mean)/std for x in X]

from keras.layers import Dense, BatchNormalization;
...
model = Sequential()
model.add(BatchNormalization())
model.add(Dense(6, input_dim=2, activation='softsign'))
model.add(BatchNormalization())
model.add(Dense(1, LeakyReLU(alpha=0.1)))

Accuracy: 0.00
[1402.37, 5.7497] => 1.1230 (expected 1.6435)
[1402.46, 6.4661] => 1.1056 (expected 0.9520)
[1402.55, 7.3768] => 1.1118 (expected 0.9508)
[1402.66, 8.2583] => 1.1365 (expected 1.3669)
[1402.78, 9.1954] => 1.0915 (expected 0.9791)
[1402.91, 10.1366] => 1.0661 (expected 0.9775)
[1402.39, 5.8727] => 1.1186 (expected 0.9725)
[1402.48, 6.7135] => 1.1042 (expected 1.1448)
[1402.58, 7.6265] => 1.1185 (expected 1.4263)
[1402.69, 8.4922] => 1.1349 (expected 0.9590)

ll, accuracy = model.evaluate(X, predictions)

# first neural network with keras make predictions
from numpy import loadtxt
from keras.models import Sequential
from keras.layers import Dense, BatchNormalization
from keras.layers import LeakyReLU
 
# load the dataset
dataset = loadtxt('KQ.txt', delimiter=',')
 
# split into input (X) and output (y) variables
X = dataset[:,0:2]
y = dataset[:,2]
 
# define the keras model
model = Sequential()
model.add(BatchNormalization())
model.add(Dense(6, input_dim=2, activation='softsign'))
model.add(BatchNormalization())
model.add(Dense(1, LeakyReLU(alpha=0.1)))
 
# compile the keras model
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['accuracy'])
 
# fit the keras model on the dataset
model.fit(X, y, epochs=150, batch_size=10, verbose=0)
 
# make class predictions with the model
predictions = model.predict(X)
 
# evaluate the keras model
ll, accuracy = model.evaluate(X, predictions)
print('Loss: %.2f' % ll)
print('Accuracy: %.2f' % (accuracy*100))
 
# summarize the first 5 cases
for i in range(10):
    print('%s => %0.4f (expected %0.4f)' % (X[i].tolist(), predictions[i], y[i]))

print('Accuracy: %.20f' % (accuracy*1000000))