na.actions: Comprehensive Library For Handling Missing Values

na.actions is a comprehensive library for handling missing (NA) values. laregly extending the existing na.* functions found in the stats packages. These Functions provide for detection, removal, replacement, recollection, imputation, etc. of NAs. In short, it is a comprehensive tool set for common operations when working with missing values.

Feature List

• Extents functionality of of existing na.* functions found in the stats package.
• NA replacement
  • type/class and length-safe replacement
  • replace by scalar, vector, functions
  • imputation with complex formula and methods
  • recall/track which value have been replaced
  • high-level function for common replacements: mean, median, max, min, zero, Inf

Installation

Github

library(devtools)
install_github( "decisionpatterns/na.actions")

CRAN

Coming Soon ...

Function List

Detections

• which_na
• all_na
• any_na

Removal / Ommision / Exclusion

• na_rm
• na_trim

Replacement / Imputation

There are a num

Constants

• na_zero - 0
• na.inf / na.neginf - Inf/-Inf
• na_constant

Computed (univariate):

• na_n - count of NA values
• na_max - maximum
• na_min - minumum
• na_mean - mean
• na_median - median value

Computed (univariate, ordered/time-series objects):

• na_loess - loess smoother, assumes values are ordered
• na_spline -
• na_locf - last observation carried forward, assumes ordered
• na_nocb - next observation carried backwards, assumes ordered
• na_structTS - Kalman Filter replacement

Computed (multivariate)

• na_predict - use a model
• na_roughfix -
• na_approx -

General:

• na_replace - atomic vectors only
• na_explicit - atomic or recursive structures

Miscellaneous

• na.roughfix.POSIXct - use with datetime values

Examples

na.replace( c(1,NA,3), 2)