cudia (version 0.1)

cudia: CUDIA: cross-level imputation framework

Description

Estimate the CUDIA model parameters, then output cross-level imputed values. The default algorithm is set to the Bregman deterministic clustering algorithm in the referenced paper. Currently, only Gaussian-type data are supported.

Usage

cudia(formula, data, K, ...)

Arguments

formula

a symbolic description of the model to be fit. e.g. x~y+z means that the aggregate-level summary x is cross-level imputed using individual-level data y and z.

data

a data frame object in the model.

K

a number of intrinsic clusters.

...

other algorithm operational parameters

Value

An object of class cudia, basically a list including elements

indiv

original individual-level data

fitted.values

cross-level imputed aggregated data

theta

parameter vectors for individual-level clusters

eta

a parameter vector for aggregate-level clusters

Nk

estimated cluster sizes

xlab

variable names of individual-level data

References

Y. Park and J. Ghosh, CUDIA: Probabilistic Cross-level Imputation using Individual Auxiliary Information, ACM Trans-IST, 2012.

See Also

print, plot methods

Examples

data(cudia_simul,package="cudia")
mod.sim <- cudia(aggr~indiv,cudia_simul,K=3)
summary(mod.sim)