Check your outliers! An introduction to identifying statistical outliers in R with easystats.

Area of Science:

• Statistics
• Data Science
• Computational Statistics

Background:

• Keeping up with best practices for outlier diagnosis and treatment is challenging.
• Implementing recommended outlier detection methods mathematically poses additional difficulties.

Purpose of the Study:

• To provide an overview of current recommendations and best practices for outlier detection.
• To demonstrate easy and convenient implementation of these methods in R using the {performance} package.
• To cover univariate, multivariate, and model-based outlier detection techniques.

Main Methods:

• Overview of univariate, multivariate, and model-based statistical outlier detection methods.
• Demonstration of implementation in R using the {performance} package from the easystats ecosystem.
• Discussion of recommended thresholds, standard output, and plotting methods.

Main Results:

• The {performance} package in R offers a convenient way to implement recommended outlier detection practices.
• The study covers various statistical outlier detection approaches and their practical application.
• Guidance on recommended thresholds, output interpretation, and visualization is provided.

Conclusions:

• The R {performance} package facilitates the practical application of advanced outlier detection methods.
• Transparency in handling different types of outliers (exclusion vs. winsorization) is crucial.
• This work bridges the gap between theoretical recommendations and practical statistical software implementation.