Clayton and Gumbel Copula Error

[dorcasgicuku]

Working with Gumbel and Clayton Copula with weather dataset produces the following error for some set of variables.

Clayton Error
ValueError Traceback (most recent call last)
in
14 clayton_parameters = clayton_copula_model.to_dict()
15 print("The theta value of", pairedData, "is",clayton_parameters['theta'])
---> 16 clayton_copula_function()

in clayton_copula_function()
11 clayton_data = clayton_df.to_numpy()
12 #Fit the model to paired data obtained from permutations the dataset
---> 13 clayton_copula_model.fit(clayton_data)
14 clayton_parameters = clayton_copula_model.to_dict()
15 print("The theta value of", pairedData, "is",clayton_parameters['theta'])

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in fit(self, X)
182 raise ValueError("Constant column.")
183 raise ValueError("Unable to compute tau.")
--> 184 self._compute_theta()
185
186 def to_dict(self):

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in _compute_theta(self)
163 """Compute theta, validate it and assign it to self."""
164 self.theta = self.compute_theta()
--> 165 self.check_theta()
166
167 def fit(self, X):

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in check_theta(self)
126 if (not lower <= self.theta <= upper) or (self.theta in self.invalid_thetas):
127 message = 'The computed theta value {} is out of limits for the given {} copula.'
--> 128 raise ValueError(message.format(self.theta, self.copula_type.name))
129
130 def check_fit(self):

ValueError: The computed theta value -0.6795841166551324 is out of limits for the given CLAYTON copula

Gumbel Copula

ValueError Traceback (most recent call last)
in
14 gumbel_parameters = gumbel_copula_model.to_dict()
15 print(gumbel_parameters['theta'])
---> 16 gumbel_copula_function()
17 # gumbel_copula_model.fit(arch_data)
18 # gumbel_parameters = gumbel_copula_model.to_dict()

in gumbel_copula_function()
11 gumbel_data = gumbel_df.to_numpy()
12 #Fit the model to paired data obtained from permutations the dataset
---> 13 gumbel_copula_model.fit(gumbel_data)
14 gumbel_parameters = gumbel_copula_model.to_dict()
15 print(gumbel_parameters['theta'])

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in fit(self, X)
182 raise ValueError("Constant column.")
183 raise ValueError("Unable to compute tau.")
--> 184 self._compute_theta()
185
186 def to_dict(self):

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in _compute_theta(self)
163 """Compute theta, validate it and assign it to self."""
164 self.theta = self.compute_theta()
--> 165 self.check_theta()
166
167 def fit(self, X):

~\Anaconda3\envs\copulas_env\lib\site-packages\copulas\bivariate\base.py in check_theta(self)
126 if (not lower <= self.theta <= upper) or (self.theta in self.invalid_thetas):
127 message = 'The computed theta value {} is out of limits for the given {} copula.'
--> 128 raise ValueError(message.format(self.theta, self.copula_type.name))
129
130 def check_fit(self):

ValueError: The computed theta value 0.6602079416724338 is out of limits for the given GUMBEL copula.

[arfogg]

Hello,
I am having a similar error to this - @dorcasgicuku did you come to a solution? Details of my problem are below:

Environment Details

Copulas version: 0.9.0
Python version: 3.8.8 (iPython environment in Spyder)
Operating System: Windows

Error Description

Fitting Bivariate Gumbel copula returns ValueError

Steps to reproduce

from copulas.bivariate import Bivariate

# Format data
copula_arr=np.array([x,y]).T
    
# Initialise the copula
copula=Bivariate(copula_type='gumbel')
    
# Fit the copula to the extremes
copula.fit(copula_arr)

Please note that x and y are on uniform margins as required. Please see histograms of x and y below.

Error

Traceback (most recent call last):

  File "C:\Users\admin\AppData\Local\Temp\ipykernel_15604\3537192975.py", line 1, in <module>
    run_solar_earth_bi_eva.phi_d_vs_ae()

  File "C:\Users\admin\Documents\wind_waves_akr_code\run_solar_earth_bi_eva\run_solar_earth_bi_eva.py", line 680, in phi_d_vs_ae
    bs_copula_arr.append(fit_copula_to_extremes.fit_copula_bivariate(bs_phid_um, bs_ae_um, 'phi_d', 'AE'))

  File "C:\Users\admin\Documents\wind_waves_akr_code\bi_multi_variate_eva\fit_copula_to_extremes.py", line 61, in fit_copula_bivariate
    copula.fit(copula_arr)

  File "C:\Users\admin\anaconda3\lib\site-packages\copulas\bivariate\base.py", line 186, in fit
    self._compute_theta()

  File "C:\Users\admin\anaconda3\lib\site-packages\copulas\bivariate\base.py", line 167, in _compute_theta
    self.check_theta()

  File "C:\Users\admin\anaconda3\lib\site-packages\copulas\bivariate\base.py", line 130, in check_theta
    raise ValueError(message.format(self.theta, self.copula_type.name))

ValueError: The computed theta value 0.9601028214324197 is out of limits for the given GUMBEL copula.

Experimenting, the Gumbel copula has a theta range of 1-inf:

from copulas.bivariate import Bivariate
copula=Bivariate(copula_type='gumbel')
print(copula.theta_interval)
out: [1, inf]

Given that I'm inputing data on uniform margins, why is the copula fitting the data in a way that's out of bounds? Please note that I have also used this same copula from this package without errors.

Thanks,
Alexandra

[arfogg]

Just to note I came to a solution on this. Essentially the data I was putting in to copula.fit was not related as it should be. Essentially x and y were almost independent, and hence the fitting was not a success.

Hence I'm not sure if this is a bug. If that's the case, it could be helpful to make the error message more verbose if possible!
Thanks,
Alexandra