Skip to contents

Visualize bHIVE/honeycombHIVE Results

Source: R/visualizeHIVE.R
visualizeHIVE.Rd

This function produces publication-quality visualizations for the results generated by the bHIVE or honeycombHIVE functions. Users can specify one or more layers to visualize and choose from several plot types:

Usage

visualizeHIVE(
  result,
  X = NULL,
  plot_type = c("scatter", "boxplot", "violin", "density"),
  feature = NULL,
  transform = TRUE,
  transformation_method = c("PCA", "UMAP", "tSNE", "none"),
  title = "HIVE Results",
  layer = 1,
  task = c("clustering", "classification", "regression"),
  ...
)

Arguments

result

A list object produced by bHIVE or honeycombHIVE. For multilayer models, each element represents one layer.

X

Optional. A numeric matrix or data frame of the original input features. If provided, data points will be plotted along with the prototypes.

plot_type

Character string specifying the type of plot to generate. Options are:

  • "scatter": A scatterplot of data points and prototypes.

  • "boxplot": A boxplot of a selected feature by group with prototypes overlaid.

  • "violin": A violin plot of a selected feature by group with prototypes overlaid.

  • "density": Density plots of a selected feature by group with prototype markers.

feature

Optional. For "boxplot", "violin", or "density" plots, the name or index of the feature in X to display. If NULL, the first column is used.

transform

Logical. If TRUE and the data (or prototypes) has more than two columns, the specified transformation is applied for scatterplots.

transformation_method

Character. The method used for dimensionality reduction. Options are "PCA", "UMAP", "tSNE", or "none".

title

Character. Title for the plot.

layer

Integer vector indicating which layer(s) of the result to visualize. For bHIVE outputs, default is 1. For multilayer honeycombHIVE outputs, specify one or more layer indices.

task

Character. The prediction task for the result: one of "clustering", "classification", or "regression". This is used to determine how grouping is computed.

...

Additional arguments passed to PCA, tSNE, UMAP, or ggplot functions.

Value

A ggplot object representing the visualization.

Details

  • "scatter": A scatterplot of data points and prototypes. When X has more than two columns and transform is TRUE, a dimensionality reduction method is applied.

  • "boxplot": A boxplot of a selected feature by group with prototype values overlaid.

  • "violin": A violin plot of a selected feature by group with prototype values overlaid.

  • "density": Density plots of a selected feature by group with prototype markers.

For scatterplots the transformation can be selected from "PCA", "UMAP", "tSNE", or "none". When multiple layers are visualized, the prototypes and the corresponding grouping information from each layer are combined and faceted by layer.

Examples

data(iris)
X <- as.matrix(iris[, 1:4])

# Run honeycombHIVE for clustering 
res <- honeycombHIVE(X = X, 
                    task = "clustering", 
                    epsilon = 0.05,
                    layers = 3, 
                    nAntibodies = 30, 
                    beta = 5, 
                    maxIter = 10, 
                    verbose = FALSE)

# Visualize layer 2 as a scatterplot (using membership from layer 2).
visualizeHIVE(result = res,
              X = iris[, 1:4],
              plot_type = "scatter",
              title = "Layer 2: Scatterplot",
              layer = 2,
              task = "clustering")


# For classification: assume res[[1]]$predictions holds class labels.
visualizeHIVE(result = res,
              X = iris[, 1:4],
              plot_type = "violin",
              feature = "Sepal.Width",
              title = "Sepal Width by Group (Layer 1)",
              layer = 1,
              task = "classification")
#> Warning: Groups with fewer than two datapoints have been dropped.
#> ℹ Set `drop = FALSE` to consider such groups for position adjustment purposes.
#> Warning: Groups with fewer than two datapoints have been dropped.
#> ℹ Set `drop = FALSE` to consider such groups for position adjustment purposes.
#> Warning: Groups with fewer than two datapoints have been dropped.
#> ℹ Set `drop = FALSE` to consider such groups for position adjustment purposes.


# For regression: prototype grouping is overridden to a constant.
visualizeHIVE(result = res,
              X = iris[, 1:4],
              plot_type = "density",
              feature = "Sepal.Width",
              title = "Sepal Width Density (Layer 1)",
              layer = 1,
              task = "regression")