So I am learning that maximum likelihood estimation of the parameters for a gamma distribution are biased. As far as I understand there is no guarantee in general that there exists a prior (or base rate) that would give an unbiased estimate, but I don't necessarily need a proper probability distribution on the parameter for many of my applications.
Is there a penalization specifically for the gamma distribution that removes the bias on MLE of its parameters?
Sources:
- https://math.stackexchange.com/questions/4380570/prove-that-the-mle-of-the-lambda-parameter-of-gamma-distribution-is-unbiased
- https://sci-hub.se/https://doi.org/10.1109/ITNG.2015.151
- For Gamma distribution, use MLE or MoM?
- Bias of variance/precision estimator using Gamma prior
- Bias correcting penalized maximum likelihood / maximum a posteriori estimates
