def test_case_3(data, Speed):
# test case 3: non-recurrent morning congestion
Fri, _, _, fri, _ = get_certain_dayofweek(data, Speed, 4)
fig3 = plt.figure(figsize=(15, 8))
fig3.suptitle(
'test case 3 (non-recurrent morning congestion): 12-16-2016, Friday')
ax = plt.subplot(2, 2, 1)
ax.set_title('average traffic speed on Fridays')
plt.pcolor(np.swapaxes(fri, 0, 1), cmap=my_cmap, vmin=20, vmax=70)
ax = plt.subplot(2, 2, 2)
ax.set_title('average traffic speed on the target Friday')
plt.pcolor(np.swapaxes(Speed[330, :, :], 0, 1),
cmap=my_cmap,
vmin=20,
vmax=70) #12-16-2016 Friday 6:30AM - 8:30AM
ax = plt.subplot(2, 2, 3)
ax.set_title('average traffic speed of sensor0 on Fridays')
plt.plot(fri[:, 0])
ax.set_ylim([0, 70])
ax = plt.subplot(2, 2, 4)
ax.set_title('traffic speed of sensor0 on the target Friday')
plt.plot(Speed[330, :, 0]) #12-16-2016 Friday 6:30AM - 8:30AM
ax.set_ylim([0, 70])
return fig3
def test_case_1(data, Speed):
# test case 1: recurrent morning congestion
Mon, _, _, mon, _ = get_certain_dayofweek(data, Speed, 0)
fig1 = plt.figure(figsize=(15, 8))
fig1.suptitle(
'test case 1 (recurrent morning congestion): 12-05-2016, Monday')
ax = plt.subplot(2, 2, 1)
plt.pcolor(np.swapaxes(mon, 0, 1), cmap=my_cmap, vmin=20, vmax=70)
ax.set_title('average traffic speed on Mondays')
ax = plt.subplot(2, 2, 2)
plt.pcolor(np.swapaxes(Speed[319, :, :], 0, 1),
cmap=my_cmap,
vmin=20,
vmax=70) #12-05-2016 Monday 6:30AM - 8:30AM
ax.set_title('traffic speed on the target Monday')
ax = plt.subplot(2, 2, 3)
ax.set_title('average traffic speed of sensor0 on Mondays')
plt.plot(mon[:, 0])
ax.set_ylim([0, 70])
ax = plt.subplot(2, 2, 4)
ax.set_title('traffic speed of sensor0 on the target Monday')
plt.plot(Speed[319, :, 0]) #12-05-2016 Monday 6:30AM - 8:30AM sensor 1
ax.set_ylim([0, 70])
return fig1
def Experiment3(epochs, data_train, data_test, Speed):
test_speed = Speed[data_test.index, :, :]
train_speed0, _, _, _, _ = get_certain_dayofweek(Speed, 0)
train_speed1, _, _, _, _ = get_certain_dayofweek(Speed, 1)
train_speed2, _, _, _, _ = get_certain_dayofweek(Speed, 2)
train_speed3, _, _, _, _ = get_certain_dayofweek(Speed, 3)
train_speed4, _, _, _, _ = get_certain_dayofweek(Speed, 4)
train_speed5, _, _, _, _ = get_certain_dayofweek(Speed, 5)
train_speed6, _, _, _, _ = get_certain_dayofweek(Speed, 6)
print('train_speed0.shape = ', train_speed0.shape)
print('test_speed.shape = ', test_speed.shape)
# construct model inputs
look_back = 15
mode = 'uni'
train_speed_x, train_speed_y = create_dataset(train_speed, train_speed,
look_back, mode)
test_speed_x, test_speed_y = create_dataset(test_speed, test_speed,
look_back, mode)
print('look_back = ', look_back)
print('mode = ', mode)
print('train_speed_x.shape = ', train_speed_x.shape)
print('train_speed_y.shape = ', train_speed_y.shape)
print('test_speed_x.shape = ', test_speed_x.shape)
print('test_speed_y.shape = ', test_speed_y.shape)
# visualize test data
#plt.plot(test_speed_y[0,390:510,:]) #12-19-2016 Monday 6:30AM - 8:30AM
plt.plot(test_speed_y[0, :, :]) #12-16-2016 Friday
# one day is a batch
batch_size = train_speed_x.shape[1]
# define model structure
model = Sequential()
model.add(
LSTM(32,
input_shape=(look_back, train_speed_x.shape[3]),
stateful=False,
return_sequences=True))
# model.add(Dropout(0.3))
model.add(
LSTM(32,
input_shape=(look_back, train_speed_x.shape[3]),
stateful=False))
# model.add(Dropout(0.3))
model.add(Dense(1))
model.compile(loss='mean_squared_error', optimizer='adam')
# fit model
train_x = np.reshape(train_speed_x,
(train_speed_x.shape[0] * train_speed_x.shape[1],
train_speed_x.shape[2], train_speed_x.shape[3]))
train_y = np.reshape(train_speed_y,
(train_speed_y.shape[0] * train_speed_y.shape[1],
train_speed_y.shape[2]))
history = model.fit(train_x,
train_y,
epochs=epochs,
batch_size=batch_size,
verbose=1,
shuffle=True)
return model, history
import pandas as pd
import csv
from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import MinMaxScaler
import matplotlib.pyplot as plt
from matplotlib import gridspec
from pylab import *
from lib.Function_DOW_Normalizer import get_certain_dayofweek, MinMax_Normalization, transfer_scale
from lib.Function_CreateInput import create_dataset
(Traffic, Speed, data) = pickle.load(
open(
"/Users/Shuo/study/Project-predictive_study/Speed_Pred_Stage2/speed_short_term.p",
"rb"))
train_speed, _, _, _, _ = get_certain_dayofweek(data, Speed[:334], dayofweek=0)
test_speed = train_speed[-1:]
look_back = 15
mode = 'uni'
test_speed_x, test_speed_y = create_dataset(test_speed, test_speed, look_back,
mode)
def model_score(history_object, image1):
trainScore = []
for i in range(len(train_speed_x)):
trainScore.append(
model.evaluate(train_speed_x[i, :, :, :],
train_speed_y[i, :, :],
batch_size=batch_size,