prior.control {mi} | R Documentation |
Auxiliary function as user interface for adding priors for mi
procedure.
Typically only used when calling mi
.
prior.control(augment.data=FALSE, pct.aug=10, K=1)
augment.data |
add the extra rows of data with pct.aug of the
random draws from the observed data, default= FALSE |
pct.aug |
percent of N being add into the exisitng data, where N is the number of observation of the completed data set, default is 10. |
K |
the cooling parameter, default is 1. If K > 0, the imputation will randomly switch between randomly imputing data from marginal (imputing data from the observed values) and drawing from the modeled based values. And with each iteration (s), the probability of cooling (q) decreases by number of iterations, such that q = K/s. |
If both augment.data=TRUE
and K>0
are specified, it will switch
augment.data=FALSE
and use only K>0
.
The default mi
process uses K=1
. In genereal, add priors
to the mi
process will speed the convergence of the imputation.
Yu-Sung Su yajima@stat.columbia.edu
### NOT RUN #================================================= # add priors by augmenting 10 #================================================= if(0){ data(CHAIN) IMP <- mi(CHAIN, add.priors=prior.control(augment.data=TRUE, pct.aug=10)) #================================================= # add priors by randomly drawing from the marginal #================================================= ## This is the default setting IMP <- mi(CHAIN, add.priors=prior.control(K=1)) #================================================= # use no prior #================================================= IMP <- mi(CHAIN, add.priors=prior.control(K=0)) }