MISS: Multiple Imputation Sequential Sampling
In LaplacesDemon: Complete Environment for Bayesian Inference

Description Usage Arguments Details Value Author(s) References See Also Examples

This function performs multiple imputation (MI) on a numeric matrix by sequentially sampling variables with missing values, given all other variables in the data set.

1	MISS(X, Iterations=100, Algorithm="GS", Fit=NULL, verbose=TRUE)

`X`	This required argument accepts a numeric matrix of data that contains both observed and missing values. Data set X must not have any rows or columns that are completely missing. X must not have any constants. The user must apply any data transformations appropriate for these models. Missing values are assumed to be Missing At Random (MAR).
`Iterations`	This is the number of iterations to perform sequential sampling via MCMC algorithms.
`Algorithm`	The MCMC algorithm defaults to the Gibbs Sampler (GS).
`Fit`	This optional argument accepts an object of class `miss`. When supplied, `MISS` will continue where it left off, provided the user does not change the algorithm (different methods are used with different algortihms, so model parameters will not match). In short, changing algorithms requires starting from scratch.
`verbose`	Logical. When `FALSE`, only the iteration prints to the console. When `TRUE`, which is the default, both the iteration and which variable is being imputed are printed to the console.

Imputation is a family of statistical methods for replacing missing values with estimates. Introduced by Rubin and Schenker (1986) and Rubin (1987), Multiple Imputation (MI) is a family of imputation methods that includes multiple estimates, and therefore includes variability of the estimates.

The Multiple Imputation Sequential Sampler (MISS) function performs MI by determining the type of variable and therefore the sampler for each variable, and then sequentially progresses through each variable in the data set that has missing values, updating its prediction of those missing values given all other variables in the data set each iteration.

MI is best performed within a model, where it is called full-likelihood imputation. Examples may be found in the "Examples" vignette. However, sometimes it is impractical to impute within a model when there are numerous missing values and a large number of parameters are therefore added. As an alternative, MI may be performed on the data set before the data is passed to the model, such as in the IterativeQuadrature, LaplaceApproximation, LaplacesDemon, or VariationalBayes function. This is less desirable, but MISS is available for MCMC-based MI in this case.

Missing values are initially set to column means for continuous variables, and are set to one for discrete variables.

MISS uses the following methods and MCMC algorithms:

Missing values of continuous variables are estimated with a ridge-stabilized linear regression Gibbs sampler.

Missing values of binary variables that have only 0 or 1 for values are estimated either with a binary robit (t-link logistic regression model) Gibbs sampler of Albert and Chib (1993).

Missing values of discrete variables with 3 or more (ordered or unordered) discrete values are considered continuous.

In the presence of big data, it is suggested that the user sequentially read in batches of data that are small enough to be manageable, and then apply the MISS function to each batch. Each batch should be representative of the whole, of course.

It is common for multiple imputation functions to handle variable transformations. MISS does not transform variables, but imputes what it gets. For example, if a user has a variable that should be positive only, then it is recommended here that the user log-transform the variable, pass the data set to MISS, and when finished, exponentiate both the observed and imputed values of that variable.

The CenterScale function should also be considered to speed up convergence.

It is hoped that MISS is helpful, though it is not without limitation and there are numerous alternatives outside of the LaplacesDemon package. If MISS does not fulfill the needs of the user, then the following packages are recommended: Amelia, mi, or mice. MISS emphasizes MCMC more than these alternatives, though MISS is not as extensive. When a data set does not have a simple structure, such as merely continuous or binary or unordered discrete, the LaplacesDemon function should be considered, where a user can easily specify complicated structures such as multilevel, spatial or temporal dependence, and more.

Matrix inversions are required in the Gibbs sampler. Matrix inversions become more cumbersome as the number J of variables increases.

If a large number of iterations is used, then the user may consider studying the imputations for approximate convergence with the BMK.Diagnostic function, by supplying the transpose of codeFit$Imp. In the presence of numerous missing values, say more than 100, the user may consider iterating through the study of the imputations of 100 missing values at a time.

This function returns an object of class miss that is a list with five components:

`Algorithm`	This indicates which algorithm was selected.
`Imp`	This is a M x T matrix of M missing values and T iterations that contains imputations.
`parm`	This is a list of length J for J variables, and each component of the list contains parameters associated with the prediction of missing values for that variable.
`PostMode`	This is a vector of posterior modes. If the user intends to replace missing values in a data set with only one estimate per missing values (single, not multiple imputation), then this vector contains these values.
`Type`	This is a vector of length J for J variables that indicates the type of each variable, as MISS will consider it. When `Type=1`, the variable is considered to be continuous. When `Type=2`, only two discrete values (0 and 1) were found.

Statisticat, LLC software@bayesian-inference.com

Albert, J.H. and Chib, S. (1993). "Bayesian Analysis of Binary and Polychotomous Response Data". Journal of the American Statistical Association, 88(422), p. 669–679.

Rubin, D.B. (1987). "Multiple Imputation for Nonresponse in Surveys". John Wiley and Sons: New York, NY.

Rubin, D.B. and Schenker, N. (1986). "Multiple Imputation for Interval Estimation from Simple Random Samples with Ignorable Nonresponse". Journal of the American Statistical Association, 81, p. 366–374.

ABB, BMK.Diagnostic, CenterScale, IterativeQuadrature LaplaceApproximation, LaplacesDemon, and VariationalBayes.

#library(LaplacesDemon)
### Create Data
#N <- 20 #Number of Simulated Records
#J <- 5 #Number of Simulated Variables
#pM <- 0.25 #Percent Missing
#Sigma <- as.positive.definite(matrix(runif(J*J),J,J))
#X <- rmvn(N, rep(0,J), Sigma)
#m <- sample.int(N*J, round(pM*N*J))
#X[m] <- NA
#head(X)

### Begin Multiple Imputation
#Fit <- MISS(X, Iterations=100, Algorithm="GS", verbose=TRUE)
#Fit
#summary(Fit)
#plot(Fit)
#plot(BMK.Diagnostic(t(Fit$Imp)))

### Continue Updating if Necessary
#Fit <- MISS(X, Iterations=100, Algorithm="GS", Fit, verbose=TRUE)
#summary(Fit)
#plot(Fit)
#plot(BMK.Diagnostic(t(Fit$Imp)))

### Replace Missing Values in Data Set with Posterior Modes
#Ximp <- X
#Ximp[which(is.na(X))] <- Fit$PostMode

### Original and Imputed Data Sets
#head(X)
#head(Ximp)
#summary(X)
#summary(Ximp)

### or Multiple Data Sets, say 3
#Ximp <- array(X, dim=c(nrow(X), ncol(X), 3))
#for (i in 1:3) {
#     Xi <- X
#     Xi[which(is.na(X))] <- Fit$Imp[,sample.int(ncol(Fit$Imp), 1)]
#     Ximp[,,i] <- Xi}
#head(X)
#head(Ximp[,,1])
#head(Ximp[,,2])
#head(Ximp[,,3])

#End