Python3 use fmin_l_bfgs_b to get MLE of Negative Binomial for binned data.

MLE_nbinom_binned.py

optimized_result = fmin_l_bfgs_b(log_likelihood_binned,
                 x0=initial_params,
                 # fprime=log_likelihood_deriv,
                 args=(X,),
                 approx_grad=True,
                 bounds=bounds)

MLE_params = {'size': optimized_result[0][0], 'prob': optimized_result[0][1]}