def run_model(data):
cidade = data["ibgeID"].iloc[0]
data_fil = data[columns]
data_fil = data_fil[data_fil['totalCases'] > 0]
data_fil = data_fil.sort_values(by='DayNum')
data_fil = filter_inconsistent_data(data_fil)
if (data_fil['totalCases'].max() > min_casos) and (len(data_fil) > min_dias) and (cidade > 1e6):
Ne = pops[pops['ibgeID'] == cidade][' POPULAÇÃO ESTIMADA '].to_list()[0]
if len(data_fil) >= 30:
nbetas = 3
else:
nbetas = 2
try:
model = SIR_BETAS(Ne, nproc)
model.fit_lsquares(data_fil['totalCases'].to_numpy(), data_fil['DayNum'].to_numpy(),
nbetas=nbetas, stand_error=True, nrand=nrand)
temp = create_summary(model, nbetas, cidade)
parameters = outp_par.append(temp, ignore_index = True)
results_df = pd.concat([outp_data,create_output(model, data_fil, pred_days,\
cidade, Ne)], sort=False)
return(results_df)
except:
new_temp = pd.DataFrame({"city": data_fill["city"].loc[0], "ibgeID": cidade})
global not_computed
not_computed.append(new_temp)
def run_model(data):
estado = data["state"].iloc[0]
Ne = pops[estado]
model = SIR_BETAS(Ne, nproc)
model.fit_lsquares(data['totalCases'].to_numpy(),
data['DayNum'].to_numpy(),
nbetas=nbetas,
stand_error=True,
nrand=nrand)
temp = create_summary(model, nbetas, estado)
parameters = outp_par.append(temp, ignore_index=True)
results_df = pd.concat([outp_data,create_output(model, data, pred_days,\
estado, Ne)], sort=False)
tts, Y = model.predict(coefs='LS')
return (results_df)
import matplotlib.pyplot as plt
#import PA data from Painel Covida (26/06/20)
data = pd.read_csv('chart.csv', sep=';', decimal=',')
dday = 'DateTime'
data[dday] = pd.to_datetime(data[dday], yearfirst=True)
data['DayNum'] = data[dday].dt.dayofyear
#Remove the predict only days
data = data.dropna()
nbeta_max = 4
#PARA population
N = 8.074e6
models = list()
for nbetas in range(1, nbeta_max + 1):
print(nbetas)
model = SIR_BETAS(N, 16)
model.fit_lsquares(data["PA"].to_numpy(),
t=data['DayNum'].to_numpy(),
nbetas=nbetas,
stand_error=True,
nrand=16)
model.fit_ML(data["PA"].to_numpy(),
t=data['DayNum'].to_numpy(),
nbetas=nbetas,
init=model.pos_ls)
#model.fit(data["PA"].to_numpy(), t=data['DayNum'].to_numpy(), nbetas=2)
models.append(model)
#tt, Yo = model.predict()
#%%
'totalCasesPred': [],
'residuo_quadratico': [],
'res_quad_padronizado': [],
'suscetivel': [],
'infectado': [],
'recuperado': []
})
#%%
#the main loop
for i, estado in enumerate(estados):
print(estado)
Ne = pops[estado]
data_fil = data[data['state'] == estado][columns]
data_fil = data_fil.sort_values(by='DayNum')
model = SIR_BETAS(Ne, nproc)
model.fit_lsquares(data_fil['totalCases'].to_numpy(),
data_fil['DayNum'].to_numpy(),
nbetas=nbetas,
stand_error=True,
nrand=nrand)
temp = create_summary(model, nbetas, estado)
outp_par = outp_par.append(temp, ignore_index=True)
outp_data = pd.concat([outp_data,create_output(model, data_fil, pred_days,\
estado)], sort=False)
tts, Y = model.predict(coefs='LS')
if gen_plots:
if i % 12 == 0:
plt.figure()
plt.subplot(3, 4, (i % 12) + 1)
plt.plot(model.t, model.Y, label='Data')