Introduction to Ssarkartrim Package

Overview

The Ssarkartrim package provides a robust trimmed-k mean estimator for numeric data. It is designed to reduce the influence of outliers by removing the k smallest and k largest values before computing the mean. This vignette walks through installation, usage, and interpretation of the function.

---

Installation

To install the package from source:

```r

install.packages(“Ssarkartrim_1.0.0.tar.gz”, repos = NULL, type = “source”)

library(Ssarkartrim)

Function: kTrimMean()

kTrimMean(dat, k)

dat : A numeric vector.

k : Number of smallest and largest values to trim.

Returns the trimmed-k mean of the data. If 2k >= length(dat), the function returns NA with a warning

Example 1 :

set.seed(5400)

dat <- rexp(20, rate = 0.5)

kTrimMean(dat, k = 2)

Output

[1] 1.592987

This trims the 2 smallest and 2 largest values from dat and computes the mean of the

remaining 16 values.

Comparison with Mean and Median

mean(dat)

median(dat)

kTrimMean(dat, k = 2)

This shows how kTrimMean() provides a middle ground:

mean() is sensitive to outliers.

median() is robust but may ignore distribution shape.

kTrimMean() trims extremes while preserving central tendency.

Example 2 :

small_dat <- c(1, 2, 3, 100, 200)

kTrimMean(small_dat, k = 1)

This trims the the lowest and largest values

Edge Case - Example 3:

short_dat <- c(5, 10)

kTrimMean(short_dat, k = 1)

Output

[1] NA

Warning message:

Not enough data to trim k smallest and largest values.

This shows the function’s built-in safeguard when trimming exceeds available data.

Use Cases

Robust estimation in small samples

Outlier-resistant summary statistics

Teaching robust statistics in coursework

Comparing estimators in simulation studies

Package Development Notes

This package was built using:

devtools::create()

Roxygen2 for documentation

document() to generate .Rd files

R CMD build and R CMD check –as-cran for validation

Rd2pdf to generate the manual

usethis::use_vignette() to create this vignette

Author

Shouhardyo Sarkar,

Department of Statistics and Actuarial Science,

Schaeffer Hall, Iowa City , IA 52240

The University of Iowa,USA

Email: shouhardyo-sarkar@uiowa.edu

Conclusion

The Ssarkartrim package offers a simple, reproducible, and pedagogically useful implementation of the trimmed-k mean.