PCA-Based Outlier Detection Using LOF

Performs Principal Component Analysis (PCA) on a data frame and identifies sample outliers using the Local Outlier Factor (LOF) method. Optionally returns a ggplot2 pair plot of the PCA scores coloured by LOF. Samples detected as outliers are removed from the returned data.

Usage

outlier_pca_lof(df, return_ggplot = TRUE, verbose = TRUE)

Arguments

df

A data frame with samples in rows and features in columns. Must not contain missing values.

return_ggplot

Logical; if TRUE, returns a ggplot2 pair plot highlighting outliers.

verbose

Logical; if TRUE, prints progress messages.

Value

A list with the following elements:

df

Filtered data frame with outliers removed. If missing values are present, the original input is returned unchanged.

plot_samples_outlier

A ggplot2 object showing PCA and LOF scores, or NULL if return_ggplot = FALSE or if missing values were found.

id_samples_outlier

A character vector of sample IDs (rownames) identified as outliers, or NULL if skipped.

Examples

if (FALSE) { # \dontrun{
result <- outlier_pca_lof(df = my_data)
head(result$df)
print(result$plot_samples_outlier)
result$id_samples_outlier
} # }