Imputation by the Support Vector Machine (mice-SVM)

This function performs proper multiple imputation using the Support Vector Machine (SVM) combined with bootstrapping, as proposed by Takahashi (2026). It is specifically designed for binary variables in high-dimensional data settings.

Usage

mice.impute.svm(
  y,
  ry,
  x,
  wy = NULL,
  type = NULL,
  C = 1,
  scaled = TRUE,
  kernel = "vanilladot",
  tol = 0.001,
  kpar = list(),
  ...
)

Arguments

y

Vector to be imputed

ry

Logical vector of length length(y) indicating the the subset y[ry] of elements in y to which the imputation model is fitted. The ry generally distinguishes the observed (TRUE) and missing values (FALSE) in y.

x

Numeric design matrix with length(y) rows with predictors for y. Matrix x may have no missing values.

wy

Logical vector of length length(y). A TRUE value indicates locations in y for which imputations are created.

type

A vector of length ncol(x) identifying the predictors. Captured here to avoid conflicts with the SVM 'type' argument from mice's internal calls.

C

Cost of constraints violation (default = 1).

scaled

A logical vector indicating the variables to be scaled.

kernel

The kernel function used in training and predicting (default = "vanilladot").

tol

Tolerance of termination criterion (default = 0.001).

kpar

List of hyper-parameters for the kernel function (default = list()).

...

Other named arguments to be passed to kernlab::ksvm().

Value

A vector of length sum(!ry) with imputed values.

Details

Note: When using this method in high-dimensional settings, it is recommended to set eps = 0 in the mice() call to prevent remove.lindep() from removing predictors before they reach the SVM model.

References

Takahashi, M. (2026). Multiple Imputation based on the Support Vector Machine for High-Dimensional Data with General Missing Patterns in Causal Inference. Journal of Statistical Computation and Simulation. doi:10.1080/00949655.2026.2660865

Author

Masayoshi Takahashi, 2026