net.eval()
"""
Make prediction
"""
inputs = X_all_tensor
label = Y_all_tensor
prediction = net(inputs)
label_origin = label.numpy() * Y_norm_std + Y_norm_mean
prediction_origin = prediction.detach().numpy() * Y_norm_std + Y_norm_mean
df3 = pd.DataFrame(prediction_origin, columns=['Retail and Recreation', 'Grocery and Pharmacy', 'Parks',
'Workplaces', 'Apple State Mobility Predict'])
df3['date'] = pd_all['date'].iloc[plot_index].to_numpy()
df3['State Name'] = pd_all['State Name'].iloc[plot_index].to_numpy()
df3['WeekDay'] = pd_all['WeekDay'].iloc[plot_index].to_numpy()
createFolder('./Mobility projection')
df3.to_excel('./Mobility projection/Mobility_Projection_'+model_mark+'_YYG'+YYG_projection_Date+'_MLmodel_'+ML_Model+'_'+case+isopen+'.xlsx')
PODA_Model['Google_Apple_Mobility_Projection_'+case]=df3
np.save(("./PODA_Model_"+today+".npy"), PODA_Model)
self.seq.add_module('fc_1', nn.Linear(n_feature, nodes))
self.seq.add_module('relu_1', nn.ReLU())
for i in range(layers):
self.seq.add_module('fc_' + str(i + 2), nn.Linear(nodes, nodes))
self.seq.add_module('relu_' + str(i + 2), nn.ReLU())
self.seq.add_module('fc_last', nn.Linear(nodes, n_output))
def forward(self, input):
return self.seq(input)
createFolder('./ML Files')
#nn.ReLU() -> nn.Tanh()
# load raw data
pd_all = PODA_Model['ML_Data']
# layers_number=4
# nodes_number = 20
# Assign the ML inputs
col_X_Name = [
'US Daily Confirmed', 'US Daily Confirmed Dfdt',
'US Daily Confirmed_shifted_1', 'US Daily Confirmed_shifted_3',
'US Daily Confirmed_shifted_7', 'US Daily Confirmed_shifted_10',
'US Daily Confirmed Dfdt_shifted_1', 'US Daily Confirmed Dfdt_shifted_3',
'US Daily Confirmed Dfdt_shifted_7', 'US Daily Confirmed Dfdt_shifted_10',
fig1 = plt.figure(figsize=(6, 5))
ax1 = fig1.add_subplot(1, 1, 1)
ax1.plot(x.index,
x['Google Fuel Demand Predict'],
'-',
label='Google Mobility (Predicted')
ax1.plot(x.index,
x['Apple Fuel Demand Predict'],
'--g',
label='Apple Mobility (Predicted')
ax1.plot(fuel_Demand_EIA.index - pd.DateOffset(days=day_Shift),
fuel_Demand_EIA['Gasoline'],
'--s',
label='EIA Weekly Fuel Demand')
ax1.set_xlabel('Date')
ax1.set_ylabel('Daily Motor Gasoline Demand (thousand BPD)')
ax1.set_ylim(4000, 10000)
ax1.set_title('Fuel Demand: ' + model_mark + case + ' YYG:' +
YYG_projection_Date + ' MLmodel:' + ML_Model + isopen)
ax1.legend()
# if (case == 'mean') & (isopen ==''):
# data_save.to_excel('C:/Users/hexx/Box Sync/Energy-COVID-19/Data for Website/Mobility_State_'+YYG_projection_Date+case+'.xlsx')
np.save(("./PODA_Model_" + today + ".npy"), PODA_Model)
createFolder('./PODA_Model')
copyfile('./PODA_Model_' + today + '.npy',
'./PODA_Model/PODA_Model_' + today + '.npy')
header=0)
PODA_Model['StateName_StateCode'] = df_StateName_Code
'''
Get Apple and Google Mobility Data
'''
# Get Apple mobility data
print('Read Apple Data')
'''
Need to check Apple Website to find the right url path
'''
df_Apple_Mobility = pd.read_csv(
"https://covid19-static.cdn-apple.com/covid19-mobility-data/2016HotfixDev16/v3/en-us/applemobilitytrends-"
+ Apple_File_Date + ".csv")
createFolder('./Mobility Google-Apple')
df_Apple_Mobility.to_csv('./Mobility Google-Apple/applemobilitytrends-' +
Apple_File_Date + '.csv')
apple_US = df_Apple_Mobility[
(df_Apple_Mobility['region'] == 'United States')
& (df_Apple_Mobility['transportation_type'] == 'driving')].transpose()
apple_US = apple_US.iloc[6:, ]
apple_US.rename(columns={apple_US.columns[0]: "Apple US"}, inplace=True)
apple_US.index = pd.to_datetime(apple_US.index)
############################################################################################################################
# get Google Mobility data
print('Read Google Mobility Data')
df_Google = pd.read_csv(
'https://www.gstatic.com/covid19/mobility/Global_Mobility_Report.csv?cachebust=911a386b6c9c230f',
retu = EIA_fuel['least_square'].sum()
return retu
x0 = [1, 1, 1, 1, 1, 1, 1, 1, 0.5, 0, 0]
bounds = Bounds([0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0.5, 0, 0, 0],
[1.2, 1.2, 1.2, 1.2, 1.2, 1, 1.3, 1.3, 1, 0.05, 10])
# res = minimize(min_func, x0, method='trust-constr', options={'verbose': 1, 'disp': True}, bounds=bounds)
res = minimize(min_func, x0, method='SLSQP', bounds=bounds)
print(res.x)
print(res.fun)
a = res.x
createFolder('./Fuel Demand Projection')
np.savetxt('./Fuel Demand Projection/Fuel_mobility_factor' +
google_Mobility_Day + '.csv',
a,
delimiter=",")
PODA_Model['Google_Mobility_EIA_Factor'] = a
np.save(("./PODA_Model_" + today + ".npy"), PODA_Model)
# fig1 = plt.figure(figsize=(6, 5))
# ax1 = fig1.add_subplot(1, 1, 1)
# ax1.plot(x.index, x['fuel factor']*baseline/100, '-o', label=['pred'])
# ax1.plot(fuel_Demand_EIA.index-pd.DateOffset(days=day_Shift), fuel_Demand_EIA['Gasoline']*demand_factor, '--s', label='EIA')
# ax1.set_xlabel('Date')
# ax1.set_ylabel('Y')
# ax1.set_title('fuel demand: shift:'+str(day_Shift)+' days')
from sklearn import datasets, linear_model
import matplotlib.pyplot as plt
import copy
from myFunctions import def_add_datashift, createFolder
today_x = pd.to_datetime('today')
today =today_x.strftime("%Y-%m-%d")
# today = '2020-08-07'
'''
Please download the file from https://www.covidanalytics.io/projections
'''
MIT_file_name = 'MIT_covid_analytics_projections_'+today+'.csv'
createFolder('./Mobility projection')
scenario_cases = ['mean'] #'upper', 'lower',
# shiftDay = 16 #YYG model shift between new infected and confirmed
# today_x = pd.to_datetime('today')
# today =today_x.strftime("%Y-%m-%d")
PODA_Model = np.load(("./PODA_Model_"+today+".npy"),allow_pickle='TRUE').item()
MIT_Data = pd.read_csv(MIT_file_name, header=0)
MIT_Data = MIT_Data[(MIT_Data['Country'] == 'US') &
(MIT_Data['Province']!='District of Columbia')]