MAMI: An R-package which performs model selection/averaging on multiply imputed datasets and combines the resulting estimates. The package also provides access to less frequently used model averaging techniques and offers integrated bootstrap estimation.
The package is useful if
one wants to perform model selection or model averaging on multiply imputed data and the analysis model of interest is either the linear model, the logistic model, the Poisson model, or the Cox proportional hazards model, possibly with a random intercept.
one wants to obtain bootstrap confidence intervals for model selection or model averaging estimators (with or without missing data/imputation) -- to address model selection uncertainty and to discover relationships of a small effect size.
one wants to compare different model selection and averaging techniques, easily with the same syntax.
The package is of limited use under the following circumstances:
if one is interested in model selection or averaging for models other than those listed above, for example parametric survival models, additive models, time-series analysis, and many others.
if one decides for a specific model selection or averaging technique not provided by the package, look at the manual for more details.
if the model selection/averaging problem is computationally too intensive, see Section 6.1 from the manual for more details.
Manual
The package manual can be found here.References
Schomaker, M., Heumann, C. (2014) Model Selection and Model Averaging after Multiple Imputation, Computational Statistics & Data Analysis, 71:758-770
Installation
The package can be downloaded here (do not forget to also install the dependencies).
Or simply type install.packages("MAMI", repos=c("http://R-Forge.R-project.org","http://cran.at.r-project.org"), dependencies=TRUE) in R.