These are some topics I am planning to write about.
- Mildly informative priors for regression coefficients in generalized linear models.
- Standardizing input variables.
- Encoding of categorical variables.
- Interactions and multilevel modeling.
- Priors for variances in multilevel models.
- Finding a multivariate normal approximation to the posterior for a hierarchical model, without creating impractically large Hessian or covariance matrices.
- Speeding up Stan’s warmup phase by first finding a multivariate normal approximation to the posterior.
- Optimal point estimates.
- Using cloud computing (e.g. Amazon Web Services) with Stan, OpenBUGS, and JAGS when you need more computational muscle.
- Adaptive thinning for MCMC output.
- Scale usage modeling for analyzing Likert-scale survey data.