Example data frames for missing data library

SUMMARY:

The cholesterol and cholesterolImpExample data frames have 28 rows and 3 columns. Cholesterol measurements were recorded for 28 patients treated for heart attacks at a Pennsylvania medical center.

DATA DESCRIPTION:

These data frames contain the following columns:

ARGUMENTS:

chol2
numeric, cholesterol readings 2 days after a heart attack.
chol4
numeric, cholesterol readings 4 days after a heart attack.
chol14
numeric, cholesterol readings 14 days after a heart attack.

chol14 has 9 missing values. In the cholesterolImpExample data frame this variable is an miVariable object, containing the original data and 5 sets of multiple imputations for the missing values.

SOURCE:

Ryan, B. F. and Joiner, B. L. (1994) Minitab Handbook (Third Edition). Wadsworth, Belmont, CA.

REFERENCES:

Schafer, J. L. (1997), Analysis of Incomplete Multivariate Data , Chapman & Hall, London.

Schimert, J., Schafer, J. L., Clarkson, D. B., Fraley, C., Hesterberg, T., (2000) Analyzing Data with Missing Values in S-PLUS , Insightful Corporation, Seattle, Washington (This manual is available on-line as file Missing.pdf), Chapter 9.

SEE ALSO:

, , , .

EXAMPLES:

# Estimate parameters using the EM algorithm: 
mdGauss(object = cholesterol, na.proc = "em") 
# Create multiple imputations. 
# Create overdispersed starting values using bootstrap samples of size n/2 
start <- list()  # Store starting values in a list 
set.seed(57)     # Set the random number seed for reproducibility 
for(i in 1:5) 
  start[[i]] <- paramIter(emGauss(cholesterol, 
                        subset=sample(1:28, size=14, T), 
                        prior="ml", 
                        control= list(trace= F))) 
# Generate imputations using those starting values 
cholesterolImpExample <- impGauss(cholesterol, prior="non",  
                                  start=start, 
                                  control=list(niter=50, seed=21))