Python implementation of the fixed effects estimator

Description

• Includes two methods, fit and predict. Computes and stores coefficients of regressors, intercept and unobserved heterogenity intercepts. Other stats like standard errors or p-values are not included
• Sklearn-based
  • Uses sklearn BaseEstimator and RegressorMixin as base classes
  • Compatibility with cross validation classes may not work (not tested)
• Implements FE à la Stata, basically ensuring that the sum of unobserved heterogeneity intercepts sum to 0, which eases interpretation. See details here