def main():
numpy.set_printoptions(threshold=numpy.nan)
model = Linear_Regression()
feature_hours = 1
data = read_data()
features = extract_features(data, feature_hours)
target = extract_target(data, feature_hours, pm25_row=9)
legends = []
for lr in (0.01, 0.02, 0.05, 0.1):
loss_data = model.train(features, target, epochs=5000, lr=lr)
legends.append(f'lr = {lr}')
plt.plot(loss_data)
plt.legend(legends, fontsize=16)
plt.xlabel('Epoch', fontsize=20)
plt.ylabel('Root Mean Squared Error (RMSE)', fontsize=20)
plt.xticks(fontsize=20)
plt.yticks(fontsize=20)
plt.show()
def main():
numpy.set_printoptions(threshold=numpy.nan)
data = read_data()
features = extract_features(data, feature_hours)
target = extract_target(data, feature_hours, pm25_row=9)
run('pm25_only', features[..., 9::18], target)
run('all', features, target)
def main():
numpy.set_printoptions(threshold=numpy.nan)
feature_hours = 1
data = read_data()
features = extract_features(data, feature_hours)
target = extract_target(data, feature_hours, pm25_row=9)
for regularization_term in (0, 0.01, 0.1, 10, 100, 1000):
model = Linear_Regression(regularization_term)
loss_data = model.train(features, target, epochs=10000)
print('lambda={}, loss={}'.format(regularization_term, loss_data[-1]))
inference(model, feature_hours, './data/test.csv',
f'./submission_q3_{regularization_term}.txt')
def main():
numpy.set_printoptions(threshold=numpy.nan)
model = Linear_Regression()
feature_hours = 1
data = read_data()
features = extract_features(data, feature_hours)
target = extract_target(data, feature_hours, pm25_row=9)
# cross_validation(model, features, target, k=12)
model.train(features, target)
numpy.savez('model.npz', b=model.b, w=model.W, feature_hours=feature_hours)
import numpy
from keras.models import Sequential
from keras.layers import Dense
from keras.optimizers import Adagrad
sys.path.append(os.path.dirname(os.path.dirname(__file__))) # flake8: noqa
from pm25_data import read_data, extract_features, extract_target
numpy.set_printoptions(threshold=numpy.nan)
feature_hours = 1
data = read_data()
features = extract_features(data, feature_hours)
target = extract_target(data, feature_hours, pm25_row=9)
model = Sequential()
linear_layer = Dense(units=1,
input_dim=18,
kernel_initializer='zeros',
bias_initializer='zeros')
model.add(linear_layer)
optimizer = Adagrad(lr=0.1, epsilon=0)
model.compile(loss='mse', optimizer=optimizer)
# Training
for step in range(10000):
cost = model.train_on_batch(features, target.T[0])
weights, biases = linear_layer.get_weights()
print("After %d trainings, the cost: %f" % (step, cost))