def test_input_empty():
with pytest.raises(Exception):
val = Avg.absAvg([])
assert False
def test_allpositive():
val = Avg.absAvg([1, 2, 3, 4, 5])
assert val == 3
def test_input_none():
with pytest.raises(Exception):
val = Avg.absAvg(None)
assert False
def test_allnegative():
val = Avg.absAvg([-1, -2, -3, -4, -5])
assert val == 3
def __init__(self, country, param, input_dir, output_dir, data,
model = 'Verhulst', label = 'Cases'):
"""
Constructor of the class
Parameters :
country : str
country name.
param : pd.DataFrame
DataFrame with the parameters.
input_dir : str
input directory.
output_dir : str
output directory.
data : pd.DataFrame
DataFrame with the historical data.
data_der : pd.DataFrame
DataFrame with the historical data derivative. The default is
'None'.
model : str, optional
str to identify the model used: possible values :
{'Verhulst', 'Gompertz', 'Mixed_VG', 'Allee'}.
The default is 'Verhulst'.
label : str, optional
str in {'Cases', 'Deads'}. The default is 'Cases'.
Returns :
None.
"""
self.__country__ = country
line = param[param['country'] == country]
self.__line__ = line
self.__total__ = int(line['total_n'])
self.__err__ = float(line['err_real_data'])
self.__output_dir__ = output_dir
self.__data__ = data
if model == 'Verhulst':
self.__mod__ = Verhulst(data[self.__country__])
elif model == 'Gompertz':
self.__mod__ = Gompertz(data[self.__country__])
elif model == 'Mixed_VG':
self.__mod__ = Mixed_VG(data[self.__country__])
elif model == 'Allee':
self.__mod__ = Allee(data[self.__country__], None)
elif model == 'AVG':
self.__mod__ = Avg(data[self.__country__], None)
self.__label__ = label
self.__stop_day__ = stop_day
if label == 'Cases':
self.__stop_day__datum = pd.read_csv(path_join(input_dir,
'stop_day_cases.csv'),
sep = ';')[self.__country__]
elif label == 'Deads':
self.__stop_day__datum = pd.read_csv(path_join(input_dir,
'stop_day_deads.csv'),
sep = ';')[self.__country__]
class Covid_predictor():
def __init__(self, country, param, input_dir, output_dir, data,
model = 'Verhulst', label = 'Cases'):
"""
Constructor of the class
Parameters :
country : str
country name.
param : pd.DataFrame
DataFrame with the parameters.
input_dir : str
input directory.
output_dir : str
output directory.
data : pd.DataFrame
DataFrame with the historical data.
data_der : pd.DataFrame
DataFrame with the historical data derivative. The default is
'None'.
model : str, optional
str to identify the model used: possible values :
{'Verhulst', 'Gompertz', 'Mixed_VG', 'Allee'}.
The default is 'Verhulst'.
label : str, optional
str in {'Cases', 'Deads'}. The default is 'Cases'.
Returns :
None.
"""
self.__country__ = country
line = param[param['country'] == country]
self.__line__ = line
self.__total__ = int(line['total_n'])
self.__err__ = float(line['err_real_data'])
self.__output_dir__ = output_dir
self.__data__ = data
if model == 'Verhulst':
self.__mod__ = Verhulst(data[self.__country__])
elif model == 'Gompertz':
self.__mod__ = Gompertz(data[self.__country__])
elif model == 'Mixed_VG':
self.__mod__ = Mixed_VG(data[self.__country__])
elif model == 'Allee':
self.__mod__ = Allee(data[self.__country__], None)
elif model == 'AVG':
self.__mod__ = Avg(data[self.__country__], None)
self.__label__ = label
self.__stop_day__ = stop_day
if label == 'Cases':
self.__stop_day__datum = pd.read_csv(path_join(input_dir,
'stop_day_cases.csv'),
sep = ';')[self.__country__]
elif label == 'Deads':
self.__stop_day__datum = pd.read_csv(path_join(input_dir,
'stop_day_deads.csv'),
sep = ';')[self.__country__]
def stop_day_value(self):
"""
This method returns the real datum for stop_day
Returns :
int64
Real datum for stop_day.
"""
return self.__stop_day__datum
def stop_day_rel_err(self):
"""
This method returns the real datum for stop_day
Returns :
int64
Real datum for stop_day.
"""
pred = self.predict_until(self.__stop_day__, 'stop_day_pred',
do_plt = False)
scaled_pred = self.__total__ * float(pred[-1])
return abs(
self.__stop_day__datum - scaled_pred) / self.__stop_day__datum
def predict_until_n(self, prediction_days, date_to_predict,
name = 'pred', do_plt = True):
"""
Predict the future outcome until the given date using historical data
Parameters :
prediction_days : int
int of the day to predict.
date_to_predict : str
str with the ending day of the prevision.
name : str
str with the name for the graph. The default is 'pred'.
do_plt : bool
if True the plot will be done. The default is True
Returns :
pred: list
scaled result of the prediction from the first day to
date_to_predict
"""
pred = self.__mod__.prediction_series(self.__line__,
limit = prediction_days)
if do_plt:
plt_str = self.__country__ + "_" + self.__label__ + name
plot_prediction(self.__data__['date'],
self.__data__[self.__country__], pred,
self.__output_dir__, err = self.__err__,
index_2 = range(len(pred)), name = plt_str)
return pred
def predict_until(self, date_to_predict, name = 'pred', do_plt = True):
"""
Predict the future outcome until the given date using historical data
Parameters :
date_to_predict : str
str with the ending day of the prevision.
name : str
str with the name for the graph. The default is 'pred'.
do_plt : bool
if True the plot will be done. The default is True
Returns :
pred: list
scaled result of the prediction from the first day to
date_to_predict
"""
prediction_date = datetime.strptime(date_to_predict,
hour_format)
prediction_days = (prediction_date - initial_day).days
return self.predict_until_n(prediction_days, date_to_predict, name,
do_plt)
def current_total(self):
"""
This method returns the current total
Returns :
int
current total.
"""
return self.__total__
def predict_total(self):
"""
This method returns the predicted total
Returns :
float
predicted total.
"""
return self.__total__ * self.__mod__.predict_max(self.__line__)
def error_margin(self):
"""
This method returns the predicted total
Returns :
float
predicted total.
"""
return self.__total__ * self.__err__
def param_table(self, data, data_der=None):
"""
Solve the optimization problem of the model for all the countries
returning a dataframe with the results of the simulation
Parameters :
data : pd.DataFrame/Series
data to be used.
data_der : pd.DataFrame
DataFrame with the historical data derivative. The default is
'None'.
Returns :
param_data : pd.DataFrame
dataframe of parameters.
"""
param_data = pd.DataFrame(columns=self.__new_cols__)
for col in data.columns:
if col != 'date' and col not in self.__warning_list__:
col_len = len(data[col])
useful_data = col_len - first_day(data[col])
if self.__model__ == 'Verhulst':
mod = Verhulst(data[col])
elif self.__model__ == 'Gompertz':
mod = Gompertz(data[col])
elif self.__model__ == 'Mixed_VG':
mod = Mixed_VG(data[col])
elif self.__model__ == 'Allee':
mod = Allee(data[col], data_der[col])
elif self.__model__ == 'AVG':
mod = Avg(data[col], data_der[col])
self.__param_cols__ = mod.param_cols
self.__initial_p__ = mod.initial_p
self.__model_bounds__ = mod.model_bounds
#at the first time we have to add the other columns
if len(param_data.columns) == len(self.__new_cols__):
param_data = pd.DataFrame(columns=self.__new_cols__ +
self.__param_cols__)
'''
res = minimize(fun = mod.MSE, x0 = self.__initial_p__,
args = (data[col]), jac = grad_MSE,
method = 'L-BFGS-B',
bounds = self.__model_bounds__)
'''
with warnings.catch_warnings(record=True) as w:
res = minimize(fun=mod.MSE,
x0=self.__initial_p__,
method='L-BFGS-B',
bounds=self.__model_bounds__)
if len(w) > 0:
self.__warning_list__.append(col)
else:
err = sqrt(res.fun)
line = [
col, useful_data,
err * sqrt(col_len / float(useful_data))
]
for i in range(len(self.__param_cols__)):
line.append(res.x[i])
param_data.loc[len(param_data)] = line
if not res.success:
print(self.__model__, col, "unsuccess")
return param_data