Comprehensive robust correlation analysis with multiple correlation methods, outlier detection, diagnostic plots, and bootstrap confidence intervals. Includes Spearman, Kendall, Percentage Bend, Biweight Midcorrelation, Minimum Volume Ellipsoid (MVE), and Minimum Covariance Determinant (MCD) correlation methods.
Usage
robustcorrelation(
data,
dep,
method = "spearman",
matrix_type = "upper",
show_pvalues = TRUE,
sig_level = 0.05,
p_adjust_method = "none",
outlier_detection = FALSE,
outlier_method = "mcd",
outlier_threshold = 2.5,
bootstrap_ci = FALSE,
n_bootstrap = 1000,
confidence_level = 0.95,
show_heatmap = TRUE,
heatmap_colors = "blue_white_red",
low_color = "#3B82C5",
mid_color = "white",
high_color = "#E74C3C",
show_diagnostics = FALSE,
plot_width = 600,
plot_height = 450,
decimal_places = 3
)Arguments
- data
The data as a data frame.
- dep
List of continuous variables for robust correlation analysis. All variables must be numeric with sufficient variation.
- method
Robust correlation method to use. Spearman and Kendall are rank-based, percentage bend and biweight are robust to outliers, MVE and MCD are based on robust covariance estimation.
- matrix_type
Display type for correlation matrix.
- show_pvalues
Display significance p-values in correlation matrix.
- sig_level
Significance level for correlation tests.
- p_adjust_method
Method for adjusting p-values for multiple testing.
- outlier_detection
Perform outlier detection analysis.
- outlier_method
Method for detecting outliers in the data.
- outlier_threshold
Threshold for outlier detection (in standard deviations or equivalent).
- bootstrap_ci
Calculate bootstrap confidence intervals for correlations.
- n_bootstrap
Number of bootstrap iterations for confidence intervals.
- confidence_level
Confidence level for bootstrap intervals.
- show_heatmap
Display correlation matrix as a heatmap visualization.
- heatmap_colors
Color scheme for correlation heatmap.
- low_color
Color for low (negative) correlation values (when using custom colors).
- mid_color
Color for mid (zero) correlation values (when using custom colors).
- high_color
Color for high (positive) correlation values (when using custom colors).
- show_diagnostics
Generate diagnostic plots for robust correlation analysis.
- plot_width
Width of plots in pixels.
- plot_height
Height of plots in pixels.
- decimal_places
Number of decimal places for correlation coefficients.
Value
A results object containing:
results$instructions | a html | ||||
results$package_status | a html | ||||
results$correlation_table | Correlation coefficients and significance tests | ||||
results$bootstrap_table | Bootstrap confidence intervals for correlations | ||||
results$outlier_table | Identified outliers and their statistics | ||||
results$correlation_heatmap | an image | ||||
results$outlier_plot | an image | ||||
results$diagnostic_plots | an image | ||||
results$scatterplot_matrix | an image |
Tables can be converted to data frames with asDF or as.data.frame. For example:
results$correlation_table$asDF
as.data.frame(results$correlation_table)
Examples
# \donttest{
# Load test data
data("mtcars")
# Basic robust correlation analysis
robustcorrelation(
data = mtcars,
dep = c("mpg", "hp", "wt", "qsec"),
method = "spearman"
)
# Advanced robust correlation with outlier detection
robustcorrelation(
data = mtcars,
dep = c("mpg", "hp", "wt", "qsec", "disp"),
method = "percentage_bend",
outlier_detection = TRUE,
outlier_method = "mcd",
bootstrap_ci = TRUE,
n_bootstrap = 1000
)
# Multiple robust methods comparison
robustcorrelation(
data = mtcars,
dep = c("mpg", "hp", "wt"),
method = "biweight",
show_heatmap = TRUE,
matrix_type = "lower"
)
# }