def test(model1,mains,appliances,op):
begin=int(input("Enter start point for testing ==> "))
end=int(input("Enter end point for testing ==> "))
X=np.array(mains.iloc[begin:end])
y=np.array(appliances[begin:end])
y=y.reshape((len(X),op))
print(X.shape)
pred1=model1.test(X)
pred1=threshold(pred1,0.5)
plt.subplot(2,1,1)
plt.plot(y,'b')
plt.ylabel("Actual")
plt.subplot(2,1,2)
plt.plot(pred1,'r')
plt.ylabel("Predicted")
plt.xlabel("time")
plt.legend()
plt.savefig(get_current_dir()+'/Outputs/'+model1.Model_name+".png")
plt.close()
print("********************************")
print("********************************")
print("Confusion Matrix for "+model1.Model_name)
print(confusion_matrix(y[:,0],pred1[:,0]))
print("********************************")
print("Accuracy Score for "+model1.Model_name)
print(accuracy_score(y[:,0],pred1[:,0]))
print("********************************")
def create_model(self):
model = Sequential()
# 1D Conv
model.add(
Conv1D(10,
3,
input_shape=(5, 1),
padding="same",
strides=1,
activation='relu'))
model.add(MaxPooling1D(pool_size=2, strides=None, padding='valid'))
#Bi-directional LSTMs
model.add(
Bidirectional(LSTM(128, return_sequences=True, stateful=False),
merge_mode='concat'))
model.add(
Bidirectional(LSTM(256, return_sequences=False, stateful=False),
merge_mode='concat'))
model.add(Dropout(0.05))
# Fully Connected Layers
model.add(Dense(128, activation='tanh'))
model.add(Dense(self.output_dim, activation='sigmoid'))
# model.add(ThresholdedReLU(theta=0.5))
model.compile(loss='binary_crossentropy', optimizer='adam')
plot_model(model,
to_file=get_current_dir() + '/Models/' + self.Model_name +
'.png',
show_shapes=True)
return model
def create_model(self):
model = Sequential()
# 1D Conv
model.add(
Conv1D(16,
4,
activation="relu",
padding="same",
strides=1,
input_shape=(4, 1)))
model.add(Conv1D(8, 4, activation="relu", padding="same", strides=1))
# Bi-directional LSTMs
model.add(
Bidirectional(GRU(64, return_sequences=True, stateful=False),
merge_mode='concat'))
model.add(
Bidirectional(GRU(256, return_sequences=False, stateful=False),
merge_mode='concat'))
# model.add(Dropout(0.1))
# Fully Connected Layers
model.add(Dense(128, activation='relu'))
model.add(Dense(self.output_dim, activation='sigmoid'))
# model.add(ThresholdedReLU(theta=0.5))
model.compile(loss='mse', optimizer='adam')
plot_model(model,
to_file=get_current_dir() + '/Models/' + self.Model_name +
'.png',
show_shapes=True)
return model
def create_model(self):
model = Sequential()
# 1D Conv
model.add(
Conv1D(8,
4,
activation="linear",
input_shape=(1, 1),
padding="same",
strides=1))
model.add(Flatten())
# Fully Connected Layers
model.add(Dropout(0.2))
# model.add(Dense((self.batchsize-0)*8, activation='relu'))
model.add(Dense((1 - 0) * 8, activation='relu'))
model.add(Dropout(0.2))
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.2))
# model.add(Dense((self.batchsize-0)*8, activation='relu'))
model.add(Dense((1 - 0) * 8, activation='relu'))
model.add(Dropout(0.2))
# 1D Conv
# model.add(Reshape(((self.batchsize-0), 8)))
model.add(Reshape(((1 - 0), 8)))
model.add(
Conv1D(self.output_dim,
4,
activation="linear",
padding="same",
strides=1))
model.compile(loss='mse', optimizer='adam')
plot_model(model,
to_file=get_current_dir() + '/Models/' + self.Model_name +
'.png',
show_shapes=True)
return model
def exists(self):
if (os.path.isfile(get_current_dir() + '/Models/' + self.Model_name +
'.h5')):
return True
else:
return False
def load(self, name, dir=get_current_dir()):
self.model = load_model(str(dir + "/Models/" + name + ".h5"))
return self
def save(self):
self.model.save(get_current_dir() + '/Models/' + self.Model_name +
'.h5')
def load(self, name, dir=get_current_dir()):
self.model = load_model(str(dir + '/Models/' + name + '.h5'))
return self