Y_indeces)
training_model = NeuralNetwork(X_training, Y_training,
hidden_layers_neurons, alfa,
_lambda)
training_model.fit(iterations)
validation_model = NeuralNetwork(X_validation,
Y_validation,
hidden_layers_neurons,
alfa, _lambda)
validation_model.network = training_model.network
validation_model.bias_weights = training_model.bias_weights
total_RMSE += validation_model.RMSE()
total_MSE += validation_model.MSE()
total_MAE += validation_model.MAE()
total_squaredR += validation_model.squaredR()
total_RMSE = total_RMSE / len(datasets)
total_MSE = total_MSE / len(datasets)
total_MAE = total_MAE / len(datasets)
total_squaredR = total_squaredR / len(datasets)
print('lambda', _lambda, 'alfa', alfa, 'nuerons',
hidden_layers_neurons)
print('RMSE ', total_RMSE)
print('MSE ', total_MSE)
print('MAE ', total_MAE)
print('total_squaredR', total_squaredR.item((0, 0)))
print('\n\n')
params = pickle.load(infile)
w = params[0]
norm = params[1]
normFeatures = norm[0]
normLabels = norm[1]
test_path = conf.TEST_PATH
s = str(
input("Enter test file path or enter x to use \"" + test_path +
"\": "))
if (s != "x"):
test_path = s
print("Using test path: " + test_path)
data = Data(test_path, normFeatures, normLabels)
nn = NeuralNetwork(data.structure, conf.sigmoid, conf.sigmoidDeriv, w)
print("MSE On test data = " + str(nn.MSE(data.data)))
while (True):
s = str(input("Do you want to test one more example [y/n]: "))
if (s == "n"):
break
inp = [1]
for i in range(data.inSize - 1):
inp.append(float(input("Enter iput for feature #" + str(i) +
": ")))
x = np.array(inp)
x -= normFeatures[0]
x /= normFeatures[1]
y = nn.forward(x)
y *= normLabels[1]
