impute_simple: Simple imputation

In antonvsdata/amp: Smoothing Metabolomics Analyses

Description

Impute missing values using a simple imputation strategy. All missing values of a feature are imputed with the same value. It is possible to only impute features with a large number of missing values this way. This can be useful for using this function before random forest imputation to speed things up. The imputation strategies available are:

Usage

impute_simple(object, value, na_limit = 0)

Arguments

object	a MetaboSet object
value	the value used for imputation, either a numeric or one of "min", "half_min", "small_random", see above
na_limit	only impute features with the proportion of NAs over this limit. For example, if na_limit = 0.5, only features with at least half of the values missing are imputed.

Details

• a numeric value: impute all missing values in all features with the same value, e.g. 1

• "min": impute missing values of a feature with the minimum observed value of that feature

• "half_min": impute missing values of a feature with half the minimum observed value of that feature

• "small_random": impute missing values of a feature with random numbers between 0 and the minimum of that feature (uniform distribution, remember to set the seed number!).

Examples

missing <- mark_nas(merged_sample, 0)
imputed <- impute_simple(missing, value = "min")

antonvsdata/amp documentation built on Jan. 8, 2020, 3:15 a.m.