print(start_time)
# %% enter end_time
num_times = 180
end_time = num_times + start_time
print(end_time)
# %% Set the initial value for model
i0 = to_float(infected[start_time])
r0 = 0.
s0 = to_float(population) - i0
d0 = to_float(deaths[start_time])
#num_times = len(deaths)
# %%
infected = infected[start_time:end_time]
deaths = deaths[start_time:end_time]
recovered = recovered[start_time:end_time]
infected = to_float_vec(infected)
deaths = to_float_vec(deaths)
recovered = to_float_vec(recovered)
# %% Set the training data
x_train = to_float_vec(np.array([infected, deaths]).transpose())
x_trains = to_float_vec(np.array([x_train]))
# %% Solve for the initial parameters
#%%
def sigmoid(x):
return 1. / (1 + np.exp(-x))
def logit(x):
return np.log(x / (1 - x))
#%%
print(start_time)
# %% enter end_time
num_times = 180
end_time = num_times + start_time
print(end_time)
# %% Set the initial value for model
i0 = to_float(infected[start_time])
r0 = 0.
s0 = to_float(population) - i0
d0 = to_float(deaths[start_time])
#num_times = len(deaths)
# %%
infected = infected[start_time:end_time]
deaths = deaths[start_time:end_time]
recovered = recovered[start_time:end_time]
infected = to_float_vec(infected)
deaths = to_float_vec(deaths)
recovered = to_float_vec(recovered)
#%%
def sigmoid(x):
return 1. / (1 + np.exp(-x))
def logit(x):
return np.log(x / (1 - x))
#%%
def minf(x):
#num_times = len(deaths)
# %% load the trained model
mypath = './models/model_' + country_code +'experiment' +'_180'
model= tf.keras.models.load_model(mypath)
#%% make dir
if os.name != 'nt':
if not os.path.exists(r'./img/' + mypath):
os.makedirs(r'./img/' + mypath)
elif not os.path.exists(r'.\\img\\' + mypath):
os.makedirs(r'.\\img\\' + mypath)
# %% find the model num_times
#layer = model.layers[-1]
num_times = 180
end_time = start_time+num_times
#%% Set the training data
x_train_old = to_float_vec(np.array([infected,deaths]).transpose())
x_trains_old = to_float_vec(np.array([x_train_old]))
x_train = to_float_vec(np.ones((1)))
x_trains = to_float_vec(np.array([x_train]))
# %% define the SIRD solver
def solve_SIRD_discrete(num_times,beta_t,gamma_t,mu_t,s0,i0,r0,d0):
S = []
I = []
R = []
D = []
S.append(s0)
I.append(i0)
R.append(r0)
D.append(d0)
num_times = len(beta_t)
# %% load the trained model
mypath = './models/model_' + country_code + '_180'
model = tf.keras.models.load_model(mypath)
#%% make dir
if os.name != 'nt':
if not os.path.exists(r'./img/' + mypath):
os.makedirs(r'./img/' + mypath)
elif not os.path.exists(r'.\\img\\' + mypath):
os.makedirs(r'.\\img\\' + mypath)
# %% find the model num_times
layer = model.layers[-1]
num_times = layer.output_shape[1]
end_time = start_time + num_times
#%% Set the training data
x_train = to_float_vec(
np.array([infected[start_time:end_time],
deaths[start_time:end_time]]).transpose())
x_trains = to_float_vec(np.array([x_train]))
# %% define the SIRD solver
def solve_SIRD_discrete(num_times, beta_t, gamma_t, mu_t, s0, i0, r0, d0):
S = []
I = []
R = []
D = []
S.append(s0)
I.append(i0)
R.append(r0)
D.append(d0)
num_times = len(beta_t)
num_times = int(input('Enter the days to train:'))
end_time = num_times + start_time
print(end_time)
# %% Set the initial value for model
i0 = to_float(infected[start_time])
r0 = 0.
s0 = to_float(population) - i0
d0 = to_float(deaths[start_time])
#num_times = len(deaths)
# %%
infected = infected[start_time:end_time]
deaths = deaths[start_time:end_time]
recovered = recovered[start_time:end_time]
# %% Set the training data
x_train = to_float_vec(np.array([infected, deaths]).transpose())
x_trains = to_float_vec(np.array([x_train]))
# %% Solve for the initial parameters
def objective(trial):
beta = trial.suggest_uniform('beta', 0, 1)
gamma = trial.suggest_uniform('gamma', 0, 1)
mu = trial.suggest_uniform('mu', 0, 1)
S = []
I = []
R = []
D = []
S.append(s0)
#%% initialize a matrix for taining sets
import numpy as np
x_test = []
y_test = []
x_train = []
y_train = []
# %% create the dataset
S = df_confirmed.keys()
import random
test_size = 0.33
S_test = random.sample(list(S), int(test_size * len(S)))
S_train = list(set(S) - set(S_test))
for s in S_test:
country_code = s
infected = to_float_vec(df_infected[country_code].values)
deaths = to_float_vec(df_deaths[country_code].values)
recovered = to_float_vec(df_recovered[country_code].values)
population = df_information[df_information.country ==
country_code].population.values[0]
#%% set time vairables
num_times = len(infected)
# %% generate
observe_days = 30
predict_days = 1
days = observe_days + predict_days
for i in range(num_times - days + 1):
x_test.append(
np.concatenate([
infected[i:observe_days + i],