I'm using statsmodels to construct a Generalized Estimating Equation (GEE) model to compute odds ratios like so:
va = sm.cov_struct.Autoregressive()
model = sm.GEE(x, y, group)
result = model.fit()
print(result.summary())
>>> GEE Regression Results
===========================================================================================
Dep. Variable: x No. Observations: 10000
Model: GEE No. clusters: 7000
Method: Generalized Min. cluster size: 1
Estimating Equations Max. cluster size: 10
Family: Gaussian Mean cluster size: 1.4
Dependence structure: Independence Num. iterations: 2
Date: Fri, 30 May 2025 Scale: 0.112
Covariance type: robust Time: 01:46:29
================================================================================
coef std err z P>|z| [0.025 0.975]
--------------------------------------------------------------------------------
y 0.0772 0.002 31.899 0.000 0.072 0.082
==============================================================================
Skew: 2.2170 Kurtosis: 3.0326
Centered skew: 0.9448 Centered kurtosis: 10.6225
==============================================================================
I wish to get the odds ratio for x. How can I do this?
I think I should be able to use the GlobalOddsRatio but don't actually understand how to use it from the reference.
You can use result.params to find the coefficients of the fitted model.
np.exp(result.params.predictor) will give you the odds ratio.