臺灣博碩士論文加值系統

The autologistic model is commonly used for modelling spatial binary data. In
this paper, we use simulation to study the performance of the pseudo-likelihood
estimate and the Markov chain Monte Carlo maximum likelihood estimate on
this model. The asymptotic variances are also compared in the simulation for
the statistical inference. A real data set from the West Nile virus in the United States is used to illustrate these methods.

1 Introduction
2 Markov Random Field
2.1 Autologistic Regression Model
2.2 Pseudo-Likelihood Estimation
2.3 Markov Chain Monte Carlo MLE method
3 Simulation
3.1 Gibbs Sampler
3.2 Simulation Results
4 Data analysis
5 Conclusion
Reference
Appendix

[1] Bartlett, M. S. (1966). An introduction to Stochastic Processes. University
Press, Cambridge.
[2] Besag, J. E. (1972). Nearest-Neighbor Systems and the Auto-Logistic Model
for Binary Data. Journal of Royal Statistical Society B 34, 75-83
[3] Besag, J. E. (1974). Spatial Interaction and the Statistical Analysis of Lattice Systems(with discussion). Journal of Royal Statistical Society B 36, 192-236.
[4] Besag, J. E. (1975). Statistical Analysis of Non-Lattice Data. The statistician 24, 179-195.
[5] Besag, J. E. (1986). On the Statistical Analysis of Dirty Pictures (with
discussion). Journal of Royal Statistical Society B 48, 259-302.
[6] Brook, S. P. (1998). Quantitative Convergence Diagnosis for MCMC via
CUSUMS. Statistics and Computing 8, 267-274.
[7] Cressie, N. A. C. (1993). Statistics for Spatial Data. Wiley, New York.
[8] Geman, D. and Geman, S. (1984). Stochastic relaxation,Gibbs distribu-
tions,and the Bayesian reconstruction of images. IEEE Transactions on Pattern
Analysis and Machine Intelligence 6, 721-741.
[9] Geyer, C. J. (1992). Practical Markov chain Monte Carlo (with discussion).
Statistical Science 7(4), 473-511.
[10] Gidas, B. (1993). Parameter Estimation for Gibbs Distribution from Fully
Observed Data. Markov Random Fields, Academic Press.
[11] Goulard, M. Sarkka, A. and Grabarnik, P. (1996). Parameter Estimation
for Marked Gibbs Point Processes through the Maximum Pseudolikelihood
Method, Scandinavian Journal of Statistic 23, 365-379.
[12] Guyon, X. (1995). Random Fields on a Network: Modeling, Statistics and
Applications. Springer, New York.
[13] Huffer, F. W. and Wu, H. (1997). Modeling the distribution of plant species using the autologistic regression model. Environmental and Ecological Statistics 4, 49-64.
[14] Huffer, F.W. andWu, H. (1998).Markov ChainMonte Carlo for Autologistic
Regression Models with Application to the Distribution of Plant Species.
Biometrics 54, 509-524.
[15] Jensen, J. L. and MøLLer, J. (1991). Pseudolikelihood for Exponential
Family Models of Spatial Point processes. The Annals of Applied Probability 1,
445-461.
[16] Lin, P. S. and Huang, Y. W. (2006). Applications of Pseudo-likelihood Estimation on Binary Markov Random Fields. Journal of the Chinese Statistical
Association 44, 91-107.