Designation of p_obs does not affect the fit about copulae

[DanielBok]

It's cause the argument is to_pobs

import numpy as np
import pandas as pd
import scipy.stats as st

from copulae import ClaytonCopula

np.random.seed(100)

v1 = np.sort(st.norm.rvs(loc=0, scale=1.5, size=10000))
v2 = st.norm.rvs(loc=0, scale=1, size=10000)

v3 = np.sort(st.norm.rvs(loc=0, scale=1.5, size=10000))
v4 = st.norm.rvs(loc=0, scale=1, size=10000)

total_sims = pd.DataFrame([0.1 * (v1 + v2), 0.1 * (v3 + v4)], index=['V1', 'V2']).T

gumbel_ca_unif = pd.DataFrame((st.gumbel_l.cdf(total_sims['V1'], *st.gumbel_l.fit(total_sims['V1']))),
                              columns=['gumbel_V1'])
gumbel_es_unif = pd.DataFrame((st.gumbel_l.cdf(total_sims['V2'], *st.gumbel_l.fit(total_sims['V2']))),
                              columns=['gumbel_V2'])

gumbel_unif = np.asarray(pd.concat([gumbel_ca_unif, gumbel_es_unif], axis=1))

for to_pobs in [True, False]:
    cop = ClaytonCopula(2)
    cop.fit(data=gumbel_unif, to_pobs=to_pobs)

    print(f"""
to_pobs = {to_pobs}
{cop.__class__.__name__} theta: {cop.params}
{cop.__class__.__name__} Kendall : {cop.tau}")
{cop.__class__.__name__} LogLikihood : {cop.log_lik(gumbel_unif, to_pobs=to_pobs)}")
    """.strip())

[njalex22]

Ahh, dumb mistake. The results now are consistent with SAS. Thank you!