def clear_directories(self, base_path, delete_all = False):
# delete all files from a directory
print("Clearing directories...")
dir_names = self.dataloader.get_all_directory_namelist()
base_path = os.path.join(self.f_prefix, base_path)
for dir_ in dir_names:
dir_path = os.path.join(base_path, dir_)
file_names = get_all_file_names(dir_path)
if delete_all:
base_file_names = []
else:
base_file_names = self.dataloader.get_base_file_name(dir_)
[delete_file(dir_path, [file_name]) for file_name in file_names if file_name not in base_file_names]
def main():
parser = argparse.ArgumentParser()
# frame rate of video
parser.add_argument('--frame', type=int, default=1,
help='Frame of video created from plots')
# gru model
parser.add_argument('--gru', action="store_true", default=False,
help='Visualization of GRU model')
# number of validation dataset
parser.add_argument('--num_of_data', type=int, default=3,
help='Number of validation data will be visualized (If 0 is given, will work on test data mode)')
# drive support
parser.add_argument('--drive', action="store_true", default=False,
help='Use Google drive or not')
# minimum lenght of trajectory
parser.add_argument('--min_traj', type=int, default=3,
help='Min. treshold of number of frame to be removed from a sequence')
# percentage of peds will be taken for each frame
parser.add_argument('--max_ped_ratio', type=float, default=0.8,
help='Percentage of pedestrian will be illustrated in a plot for a sequence')
# maximum ped numbers
parser.add_argument('--max_target_ped', type=int, default=20,
help='Maximum number of peds in final plot')
# method to be visualized
parser.add_argument('--method', type=int, default=1,
help='Method of lstm will be used (1 = social lstm, 2 = obstacle lstm, 3 = vanilla lstm)')
# Parse the parameters
args = parser.parse_args()
prefix = ''
f_prefix = '.'
if args.drive is True:
prefix='drive/semester_project/social_lstm_final/'
f_prefix = 'drive/semester_project/social_lstm_final'
model_name = "LSTM"
method_name = get_method_name(args.method)
if args.gru:
model_name = "GRU"
plot_file_directory = 'validation'
# Directories
if args.num_of_data is 0:
plot_file_directory = 'test'
# creation of paths
save_plot_directory = os.path.join(f_prefix, 'plot',method_name, model_name,'plots/')
plot_directory = os.path.join(f_prefix, 'plot', method_name, model_name, plot_file_directory)
video_directory = os.path.join(f_prefix, 'plot',method_name, model_name,'videos/')
plot_file_name = get_all_file_names(plot_directory)
num_of_data = np.clip(args.num_of_data, 0, len(plot_file_name))
plot_file_name = random.sample(plot_file_name, num_of_data)
for file_index in range(len(plot_file_name)):
file_name = plot_file_name[file_index]
folder_name = remove_file_extention(file_name)
print("Now processing: ", file_name)
file_path = os.path.join(plot_directory, file_name)
video_save_directory = os.path.join(video_directory, folder_name)
figure_save_directory = os.path.join(save_plot_directory, folder_name)
# remove existed plots
clear_folder(video_save_directory)
clear_folder(figure_save_directory)
if not os.path.exists(video_save_directory):
os.makedirs(video_save_directory)
if not os.path.exists(figure_save_directory):
os.makedirs(figure_save_directory)
try:
f = open(file_path, 'rb')
except FileNotFoundError:
print("File not found: %s"%file_path)
continue
results = pickle.load(f)
result_arr = np.array(results)
true_trajectories = np.array(result_arr[:,0])
pred_trajectories = np.array(result_arr[:,1])
frames = np.array(result_arr[:, 4])
target_id_trajs = []
args.max_target_ped = np.clip(args.max_target_ped, 0, len(results)-1)
min_r = -10
max_r = 10
plot_offset = 1
for i in range(len(results)):
print("##########################################################################################")
name = 'sequence' + str(i).zfill(5)
print("Now processing seq: ",name)
if args.num_of_data is 0: #test data visualization
target_traj = plot_trajectories(results[i][0], results[i][1], results[i][2], results[i][3], results[i][4], name, figure_save_directory, args.min_traj ,args.max_ped_ratio, results[i][5], [min_r, max_r, plot_offset], results[i][6])
else:
target_traj = plot_trajectories(results[i][0], results[i][1], results[i][2], results[i][3],results[i][4], name, figure_save_directory, args.min_traj ,args.max_ped_ratio, results[i][5], [min_r, max_r, plot_offset], 20)
target_traj.append(results[i][2])#pedlist
target_traj.append(results[i][3])#lookup
target_id_trajs.append(target_traj)
save_video(figure_save_directory, video_save_directory, plot_file_name[file_index], args.frame)
plot_target_trajs(target_id_trajs, figure_save_directory, args.max_target_ped, plot_offset)
def main():
parser = argparse.ArgumentParser()
# frame rate of video
parser.add_argument('--frame',
type=int,
default=1,
help='Frame of video created from plots')
# gru model
parser.add_argument('--gru',
action="store_true",
default=False,
help='Visualization of GRU model')
# number of validation dataset
parser.add_argument(
'--num_of_data',
type=int,
default=0,
help=
'Number of validation data will be visualized (If 0 is given, will work on test data mode)'
)
# drive support
parser.add_argument('--drive',
action="store_true",
default=False,
help='Use Google drive or not')
# minimum lenght of trajectory
parser.add_argument(
'--min_traj',
type=int,
default=3,
help='Min. treshold of number of frame to be removed from a sequence')
# percentage of peds will be taken for each frame
parser.add_argument(
'--max_ped_ratio',
type=float,
default=0.8,
help=
'Percentage of pedestrian will be illustrated in a plot for a sequence'
)
# maximum ped numbers
parser.add_argument('--max_target_ped',
type=int,
default=20,
help='Maximum number of peds in final plot')
# method to be visualized
parser.add_argument(
'--method',
type=int,
default=1,
help=
'Method of lstm will be used (1 = social lstm, 2 = obstacle lstm, 3 = vanilla lstm)'
)
# Parse the parameters
args = parser.parse_args()
prefix = ''
f_prefix = '.'
method_name = "VANILLALSTM"
model_name = "LSTM"
if args.gru:
model_name = "GRU"
plot_file_directory = 'validation'
# Directories
if args.num_of_data is 0:
plot_file_directory = 'test'
# creation of paths
save_plot_directory = os.path.join(f_prefix, 'plot', method_name,
model_name, 'plots/')
plot_directory = os.path.join(f_prefix, 'plot', method_name, model_name,
plot_file_directory)
video_directory = os.path.join(f_prefix, 'plot', method_name, model_name,
'videos/')
plot_file_name = get_all_file_names(plot_directory)
print(plot_file_name)
for file_index in range(len(plot_file_name)):
file_name = plot_file_name[file_index]
print("Now processing: ", file_name)
file_path = os.path.join(plot_directory, file_name)
try:
f = open(file_path, 'rb')
except FileNotFoundError:
print("File not found: %s" % file_path)
continue
results = pickle.load(f)
obs_length = np.array(results[0][1])
pred_length = np.array(results[0][0])
enthalpy_max = [408, 430, 251]
enthalpy_min = [404, 422.5, 239]
enthalpy_label = [
'evaporator outlet enthalpy', 'compressor outlet enthalpy',
'condenser outlet enthalpy'
]
'''
draw enthalpy prediction diagram
'''
expected_enthalpy_arr = np.array(results[0][2])
predicted_enthalpy_arr = np.array(results[0][3])
error_enthalpy = np.zeros((expected_enthalpy_arr.shape[0], 3))
plt.figure(figsize=(10, 5))
for i in range(expected_enthalpy_arr.shape[0]):
for idx in range(3):
expected_enthalpy = expected_enthalpy_arr[i, :, idx:idx + 1]
predicted_enthalpy = predicted_enthalpy_arr[i, :, idx:idx + 1]
# predicted_enthalpy[obs_length:] = filter(predicted_enthalpy[obs_length:])
mean_obs_enthalpy = np.mean(expected_enthalpy[:obs_length])
mean_expected_enthalpy = np.mean(
expected_enthalpy[obs_length:])
mean_predicted_enthalpy = np.mean(
predicted_enthalpy[obs_length:])
# draw plot
plt.ylim(enthalpy_min[idx], enthalpy_max[idx])
plt.plot(np.arange(0, obs_length * 2, 2),
expected_enthalpy[:obs_length],
color='green',
label="Observation data",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2,
expected_enthalpy.shape[0] * 2, 2),
expected_enthalpy[obs_length:],
label="Expected " + enthalpy_label[idx],
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2,
expected_enthalpy.shape[0] * 2, 2),
predicted_enthalpy[obs_length:],
label="Predictive " + enthalpy_label[idx],
linestyle='-',
marker='+')
plt.plot([0, obs_length * 2],
[mean_obs_enthalpy, mean_obs_enthalpy],
color='black',
label="Observation length mean " +
enthalpy_label[idx])
plt.plot([obs_length * 2, expected_enthalpy.shape[0] * 2],
[mean_expected_enthalpy, mean_expected_enthalpy],
color='gray',
label="Expected mean " + enthalpy_label[idx])
plt.plot([obs_length * 2, expected_enthalpy.shape[0] * 2],
[mean_predicted_enthalpy, mean_predicted_enthalpy],
color='red',
label="Predictive mean " + enthalpy_label[idx])
plt.xlabel('Time(minutes)', fontsize=12)
plt.ylabel('Enthalpy(kJ/kg)', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/" + str(i) + "H" +
str(idx + 1) + ".png")
plt.clf()
# plt.show()
# compute error
# sc = MinMaxScaler()
# fit_data = np.concatenate((expected_enthalpy, predicted_enthalpy), axis=0)
# fit_data = sc.fit_transform(fit_data)
# expected_enthalpy = fit_data[:expected_enthalpy.shape[0]]
# predicted_enthalpy = fit_data[expected_enthalpy.shape[0]:]
# error = np.sum(np.abs(expected_enthalpy[obs_length:]-predicted_enthalpy[obs_length:]))/expected_enthalpy[obs_length:].shape[0]
error = np.abs(mean_expected_enthalpy -
mean_predicted_enthalpy)
error = np.around(error, decimals=4)
error_enthalpy[i, idx] = error
new_df = pd.DataFrame({
'h1 err': error_enthalpy[:, 0],
'h2 err': error_enthalpy[:, 1],
'h3 err': error_enthalpy[:, 2]
})
new_df.to_csv(plot_directory + "/" + "enthalpy_error_minutes" + ".csv",
index=False)
'''
draw cop prediction diagram (minutes)
'''
error_cop = np.zeros((expected_enthalpy_arr.shape[0], 1))
for i in range(expected_enthalpy_arr.shape[0]):
expected_cop_arr = (expected_enthalpy_arr[i, :, 0:1] -
expected_enthalpy_arr[i, :, 2:3]) / (
expected_enthalpy_arr[i, :, 1:2] -
expected_enthalpy_arr[i, :, 0:1])
predicted_cop_arr = (predicted_enthalpy_arr[i, :, 0:1] -
predicted_enthalpy_arr[i, :, 2:3]) / (
predicted_enthalpy_arr[i, :, 1:2] -
predicted_enthalpy_arr[i, :, 0:1])
# predicted_cop_arr[obs_length:] = filter(predicted_cop_arr[obs_length:])
mean_obs_cop = np.mean(expected_cop_arr[:obs_length])
mean_expected_cop = np.mean(expected_cop_arr[obs_length:])
mean_predicted_cop = np.mean(predicted_cop_arr[obs_length:])
plt.ylim(6, 10)
plt.plot(np.arange(0, obs_length * 2, 2),
expected_cop_arr[:obs_length],
color='green',
label="Observation data",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2, expected_cop_arr.shape[0] * 2,
2),
expected_cop_arr[obs_length:],
label="Expected COP",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2, expected_cop_arr.shape[0] * 2,
2),
predicted_cop_arr[obs_length:],
label="Predictive COP",
linestyle='-',
marker='+')
plt.plot([0, obs_length * 2], [mean_obs_cop, mean_obs_cop],
color='black',
label="Observation length mean COP")
plt.plot([obs_length * 2, expected_cop_arr.shape[0] * 2],
[mean_expected_cop, mean_expected_cop],
color='gray',
label="Expected mean COP")
plt.plot([obs_length * 2, expected_cop_arr.shape[0] * 2],
[mean_predicted_cop, mean_predicted_cop],
color='red',
label="Predictive mean COP")
plt.xlabel('Time(minutes)', fontsize=12)
plt.ylabel('COP', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/test" + str(i) + "_COP minutes.png")
plt.clf()
# plt.show()
# compute error
# sc = MinMaxScaler()
# fit_data = np.concatenate((expected_cop_arr, predicted_cop_arr), axis=0)
# fit_data = sc.fit_transform(fit_data)
# expected_cop_arr = fit_data[:expected_cop_arr.shape[0]]
# predicted_cop_arr = fit_data[expected_cop_arr.shape[0]:]
# print(expected_cop_arr.shape, predicted_cop_arr.shape)
# error = np.sum(np.abs(expected_cop_arr[obs_length:]-predicted_cop_arr[obs_length:]))/expected_cop_arr[obs_length:].shape[0]
error = np.abs(mean_expected_cop - mean_predicted_cop)
error = np.around(error, decimals=4)
error_cop[i, 0] = error
new_df = pd.DataFrame({'minutes err': error_cop[:, 0]})
new_df.to_csv(plot_directory + "/" + "COP_error_minutes.csv",
index=False)
'''
hours
'''
expected_cop_day_arr = []
predicted_cop_day_arr = []
for i in range(expected_enthalpy_arr.shape[0]):
expected_cop_arr = (expected_enthalpy_arr[i, :, 0:1] -
expected_enthalpy_arr[i, :, 2:3]) / (
expected_enthalpy_arr[i, :, 1:2] -
expected_enthalpy_arr[i, :, 0:1])
predicted_cop_arr = (predicted_enthalpy_arr[i, :, 0:1] -
predicted_enthalpy_arr[i, :, 2:3]) / (
predicted_enthalpy_arr[i, :, 1:2] -
predicted_enthalpy_arr[i, :, 0:1])
# predicted_cop_arr[obs_length:] = filter(predicted_cop_arr[obs_length:])
expected_cop_day = expected_cop_arr[obs_length:]
predicted_cop_day = predicted_cop_arr[obs_length:]
expected_cop_day = np.mean(expected_cop_day)
predicted_cop_day = np.mean(predicted_cop_day)
expected_cop_day_arr.append(expected_cop_day)
predicted_cop_day_arr.append(predicted_cop_day)
plt.ylim(6.5, 9)
hour = pred_length / 30
inveral = 7
expected_cop_day_arr = np.array(expected_cop_day_arr)
predicted_cop_day_arr = np.array(predicted_cop_day_arr)
expected_cop_day_arr = np.reshape(expected_cop_day_arr,
(expected_cop_day_arr.shape[0], 1))
predicted_cop_day_arr = np.reshape(predicted_cop_day_arr,
(predicted_cop_day_arr.shape[0], 1))
plt.xticks(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour))
plt.plot(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour),
expected_cop_day_arr,
label="Expected COP",
linestyle='--',
marker='+')
plt.plot(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour),
predicted_cop_day_arr,
label="Predictive COP",
linestyle='--',
marker='^')
plt.xlabel('Time(hours)', fontsize=12)
plt.ylabel('COP', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/" + str(hour) + "hours_COP" + ".png")
plt.clf()
# plt.show()
# compute error
# sc = MinMaxScaler()
# fit_data = np.concatenate((expected_cop_day_arr, predicted_cop_day_arr), axis=0)
# fit_data = sc.fit_transform(fit_data)
# expected = fit_data[:expected_cop_day_arr.shape[0]]
# predicted = fit_data[expected_cop_day_arr.shape[0]:]
error = np.abs(expected_cop_day_arr - predicted_cop_day_arr)
error = np.around(error, decimals=4)
print(expected_cop_day_arr.shape, predicted_cop_day_arr.shape,
error.shape)
new_df = pd.DataFrame({
'real COP': expected_cop_day_arr[:, 0],
'predict COP': predicted_cop_day_arr[:, 0],
'err': error[:, 0]
})
new_df.to_csv(plot_directory + "/" + "COP_error_hours.csv",
index=False)
def main():
parser = argparse.ArgumentParser()
# gru model
parser.add_argument('--gru',
action="store_true",
default=True,
help='Visualization of GRU model')
# Parse the parameters
args = parser.parse_args()
prefix = ''
f_prefix = '.'
method_name = "VANILLALSTM"
model_name = "LSTM"
if args.gru:
model_name = "GRU"
plot_file_directory = 'test'
# creation of paths
plot_directory = os.path.join(f_prefix, 'plot', method_name, model_name,
plot_file_directory)
plot_file_name = get_all_file_names(plot_directory)
print(plot_file_name)
for file_index in range(len(plot_file_name)):
file_name = plot_file_name[file_index]
print("Now processing: ", file_name)
file_path = os.path.join(plot_directory, file_name)
try:
f = open(file_path, 'rb')
except FileNotFoundError:
print("File not found: %s" % file_path)
continue
results = pickle.load(f)
obs_length = np.array(results[0][1])
pred_length = np.array(results[0][0])
enthalpy_max = [408, 430, 251]
enthalpy_min = [404, 422.5, 239]
enthalpy_label = [
'evaporator outlet enthalpy', 'compressor outlet enthalpy',
'condenser outlet enthalpy'
]
'''
draw enthalpy prediction diagram
'''
expected_enthalpy_arr = np.array(results[0][2])
predicted_enthalpy_arr = np.array(results[0][3])
error_enthalpy = np.zeros((expected_enthalpy_arr.shape[0], 3))
plt.figure(figsize=(10, 5))
for i in range(expected_enthalpy_arr.shape[0]):
for idx in range(3):
expected_enthalpy = expected_enthalpy_arr[i, :, idx:idx + 1]
predicted_enthalpy = predicted_enthalpy_arr[i, :, idx:idx + 1]
# predicted_enthalpy[obs_length:] = filter(predicted_enthalpy[obs_length:])
mean_obs_enthalpy = np.mean(expected_enthalpy[:obs_length])
mean_expected_enthalpy = np.mean(
expected_enthalpy[obs_length:])
mean_predicted_enthalpy = np.mean(
predicted_enthalpy[obs_length:])
# draw plot
plt.ylim(enthalpy_min[idx], enthalpy_max[idx])
plt.plot(np.arange(0, obs_length * 2, 2),
expected_enthalpy[:obs_length],
color='green',
label="Observation data",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2,
expected_enthalpy.shape[0] * 2, 2),
expected_enthalpy[obs_length:],
label="Expected " + enthalpy_label[idx],
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2,
expected_enthalpy.shape[0] * 2, 2),
predicted_enthalpy[obs_length:],
label="Predictive " + enthalpy_label[idx],
linestyle='-',
marker='+')
plt.plot([0, obs_length * 2],
[mean_obs_enthalpy, mean_obs_enthalpy],
color='black',
label="Observation length mean " +
enthalpy_label[idx])
plt.plot([obs_length * 2, expected_enthalpy.shape[0] * 2],
[mean_expected_enthalpy, mean_expected_enthalpy],
color='gray',
label="Expected mean " + enthalpy_label[idx])
plt.plot([obs_length * 2, expected_enthalpy.shape[0] * 2],
[mean_predicted_enthalpy, mean_predicted_enthalpy],
color='red',
label="Predictive mean " + enthalpy_label[idx])
plt.xlabel('Time(minutes)', fontsize=12)
plt.ylabel('Enthalpy(kJ/kg)', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/" + str(i) + "H" +
str(idx + 1) + ".png")
plt.clf()
# plt.show()
# compute error
error = np.abs(mean_expected_enthalpy -
mean_predicted_enthalpy)
error = np.around(error, decimals=4)
error_enthalpy[i, idx] = error
new_df = pd.DataFrame({
'h1 err': error_enthalpy[:, 0],
'h2 err': error_enthalpy[:, 1],
'h3 err': error_enthalpy[:, 2]
})
new_df.to_csv(plot_directory + "/" + "enthalpy_error_minutes" + ".csv",
index=False)
'''
draw cop prediction diagram (minutes)
'''
error_cop = np.zeros((expected_enthalpy_arr.shape[0], 1))
for i in range(expected_enthalpy_arr.shape[0]):
expected_cop_arr = (expected_enthalpy_arr[i, :, 0:1] -
expected_enthalpy_arr[i, :, 2:3]) / (
expected_enthalpy_arr[i, :, 1:2] -
expected_enthalpy_arr[i, :, 0:1])
predicted_cop_arr = (predicted_enthalpy_arr[i, :, 0:1] -
predicted_enthalpy_arr[i, :, 2:3]) / (
predicted_enthalpy_arr[i, :, 1:2] -
predicted_enthalpy_arr[i, :, 0:1])
# predicted_cop_arr[obs_length:] = filter(predicted_cop_arr[obs_length:])
mean_obs_cop = np.mean(expected_cop_arr[:obs_length])
mean_expected_cop = np.mean(expected_cop_arr[obs_length:])
mean_predicted_cop = np.mean(predicted_cop_arr[obs_length:])
plt.ylim(6, 10)
plt.plot(np.arange(0, obs_length * 2, 2),
expected_cop_arr[:obs_length],
color='green',
label="Observation data",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2, expected_cop_arr.shape[0] * 2,
2),
expected_cop_arr[obs_length:],
label="Expected COP",
linestyle='-',
marker='.')
plt.plot(np.arange(obs_length * 2, expected_cop_arr.shape[0] * 2,
2),
predicted_cop_arr[obs_length:],
label="Predictive COP",
linestyle='-',
marker='+')
plt.plot([0, obs_length * 2], [mean_obs_cop, mean_obs_cop],
color='black',
label="Observation length mean COP")
plt.plot([obs_length * 2, expected_cop_arr.shape[0] * 2],
[mean_expected_cop, mean_expected_cop],
color='gray',
label="Expected mean COP")
plt.plot([obs_length * 2, expected_cop_arr.shape[0] * 2],
[mean_predicted_cop, mean_predicted_cop],
color='red',
label="Predictive mean COP")
plt.xlabel('Time(minutes)', fontsize=12)
plt.ylabel('COP', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/test" + str(i) + "_COP minutes.png")
plt.clf()
# plt.show()
# compute error
error = np.abs(mean_expected_cop - mean_predicted_cop)
error = np.around(error, decimals=4)
error_cop[i, 0] = error
new_df = pd.DataFrame({'minutes err': error_cop[:, 0]})
new_df.to_csv(plot_directory + "/" + "COP_error_minutes.csv",
index=False)
'''
hours
'''
expected_cop_day_arr = []
predicted_cop_day_arr = []
for i in range(expected_enthalpy_arr.shape[0]):
expected_cop_arr = (expected_enthalpy_arr[i, :, 0:1] -
expected_enthalpy_arr[i, :, 2:3]) / (
expected_enthalpy_arr[i, :, 1:2] -
expected_enthalpy_arr[i, :, 0:1])
predicted_cop_arr = (predicted_enthalpy_arr[i, :, 0:1] -
predicted_enthalpy_arr[i, :, 2:3]) / (
predicted_enthalpy_arr[i, :, 1:2] -
predicted_enthalpy_arr[i, :, 0:1])
# predicted_cop_arr[obs_length:] = filter(predicted_cop_arr[obs_length:])
expected_cop_day = expected_cop_arr[obs_length:]
predicted_cop_day = predicted_cop_arr[obs_length:]
expected_cop_day = np.mean(expected_cop_day)
predicted_cop_day = np.mean(predicted_cop_day)
expected_cop_day_arr.append(expected_cop_day)
predicted_cop_day_arr.append(predicted_cop_day)
plt.ylim(6.5, 9)
hour = pred_length / 30
inveral = 8
expected_cop_day_arr = np.array(expected_cop_day_arr)
predicted_cop_day_arr = np.array(predicted_cop_day_arr)
expected_cop_day_arr = np.reshape(expected_cop_day_arr,
(expected_cop_day_arr.shape[0], 1))
predicted_cop_day_arr = np.reshape(predicted_cop_day_arr,
(predicted_cop_day_arr.shape[0], 1))
plt.xticks(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour))
plt.plot(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour),
expected_cop_day_arr,
label="Expected COP",
linestyle='--',
marker='+')
plt.plot(np.arange(0, expected_cop_day_arr.shape[0] * hour, hour),
predicted_cop_day_arr,
label="Predictive COP",
linestyle='--',
marker='^')
plt.xlabel('Time(hours)', fontsize=12)
plt.ylabel('COP', fontsize=12)
plt.xticks(fontsize=12)
plt.yticks(fontsize=12)
plt.legend()
plt.savefig(plot_directory + "/" + str(hour) + "hours_COP" + ".png")
plt.clf()
# plt.show()
# compute error
error = np.abs(expected_cop_day_arr - predicted_cop_day_arr)
error = np.around(error, decimals=4)
print(expected_cop_day_arr.shape, predicted_cop_day_arr.shape,
error.shape)
new_df = pd.DataFrame({
'real COP': expected_cop_day_arr[:, 0],
'predict COP': predicted_cop_day_arr[:, 0],
'err': error[:, 0]
})
new_df.to_csv(plot_directory + "/" + "COP_error_hours.csv",
index=False)