nn_impute {ggbrain}

Imputes missing values in a 2D matrix based on the nearest non-missing neighbors in a given radius

Description

Imputes missing values in a 2D matrix based on the nearest non-missing neighbors in a given radius

Arguments

in_mat

a 2D matrix to fill using nearest neighbors

neighbors

the number of closest non-NA neighboring values to return within in_mat. Default is 4.

radius

the radius (in pixels) around each missing value to search for non-missing neighbors. Default is 8.

aggfun

the function used to aggregate the neighbors in imputation. Supports "mean", "median", and "mode."

ignore_zeros

if TRUE, then zero is not a valid imputation value (since these are not data in NIfTIs)

Details

The "mode" aggfun should only be used when the matrix in_mat can be converted to integers without loss of information (i.e., the data are integerish values already).

Value

A copy of the matrix with NA values imputed by their nearest neighbors