MCMC
Units & Measurements abbreviation · Statistics
| Stands for | Markov chain Monte Carlo |
|---|---|
| Categories | Units & Measurements |
MCMC stands for Markov chain Monte Carlo, a statistics abbreviation.
A family of algorithms that draw samples from complex probability distributions by running a Markov chain whose long-run distribution is the target. It made modern Bayesian statistics practical.
Example: Posterior estimates were obtained by MCMC with four chains of 2,000 iterations each.
Usage notes
The Metropolis algorithm dates from 1953, W. K. Hastings generalized it in 1970, and Gibbs sampling spread in the late 1980s. Modern software such as Stan uses Hamiltonian Monte Carlo. Convergence is checked with diagnostics such as R-hat values close to 1 and effective sample size.
Related abbreviations
More statistics abbreviations
- MAE – Mean absolute error
- MANOVA – Multivariate analysis of variance
- MAP – Maximum a posteriori
- MAPE – Mean absolute percentage error
- MAR – Missing at random
- MCAR – Missing completely at random
- Mdn – Median
- MGF – Moment-generating function
- MLE – Maximum likelihood estimation
- MOE – Margin of error
- MSE – Mean squared error
- n – Sample size