hexbin Mark

library(mosaicr)

The hexbin mark creates hexagonal binning visualizations, perfect for displaying density in large scatter plots where individual points would overlap and create overplotting issues.

Basic Example

# Generate dense scatter plot data
set.seed(123)
n <- 1000
data <- data.frame(
  x = rnorm(n, 50, 15),
  y = rnorm(n, 50, 15)
)

spec <- list(
  plot = list(
    list(
      mark = "hexbin",
      data = list(from = "sample_data"),
      x = "x",
      y = "y",
      fill = list(count = NULL)
    )
  )
)

mosaic(spec, sample_data = data)

Customization Options

Custom Color Scheme

# Bivariate normal distribution
set.seed(456)
n <- 2000
bivar_data <- data.frame(
  x = rnorm(n, 0, 1),
  y = rnorm(n, 0, 1)
)

spec_colored <- list(
  plot = list(
    list(
      mark = "hexbin",
      data = list(from = "bivar_data"),
      x = "x",
      y = "y",
      fill = list(count = NULL),
      fillOpacity = 0.8
    )
  ),
  colorScheme = "viridis"
)

mosaic(spec_colored, bivar_data = bivar_data)

Multiple Distributions

# Create data with multiple clusters
set.seed(789)
n_per_cluster <- 500

cluster_data <- data.frame(
  x = c(
    rnorm(n_per_cluster, 20, 5),   # Cluster 1
    rnorm(n_per_cluster, 60, 8),   # Cluster 2
    rnorm(n_per_cluster, 40, 6)    # Cluster 3
  ),
  y = c(
    rnorm(n_per_cluster, 30, 7),   # Cluster 1
    rnorm(n_per_cluster, 70, 5),   # Cluster 2
    rnorm(n_per_cluster, 20, 4)    # Cluster 3
  )
)

spec_clusters <- list(
  plot = list(
    list(
      mark = "hexbin",
      data = list(from = "cluster_data"),
      x = "x",
      y = "y",
      fill = list(count = NULL)
    )
  ),
  colorScheme = "plasma"
)

mosaic(spec_clusters, cluster_data = cluster_data)

With Background Grid

# Hexbin with grid for reference
spec_grid <- list(
  plot = list(
    list(
      mark = "hexgrid",
      stroke = "#aaa",
      strokeOpacity = 0.3
    ),
    list(
      mark = "hexbin",
      data = list(from = "bivar_data"),
      x = "x",
      y = "y",
      fill = list(count = NULL),
      fillOpacity = 0.7
    )
  )
)

mosaic(spec_grid, bivar_data = bivar_data)

Real-world Example

# Using a larger dataset (simulated real-world scenario)
set.seed(999)
n <- 5000

# Simulate correlation between variables
x_vals <- rnorm(n, 100, 20)
y_vals <- 0.8 * x_vals + rnorm(n, 0, 15)

realworld_data <- data.frame(
  measurement_a = x_vals,
  measurement_b = y_vals
)

spec_realworld <- list(
  plot = list(
    list(
      mark = "hexbin",
      data = list(from = "realworld_data"),
      x = "measurement_a",
      y = "measurement_b",
      fill = list(count = NULL)
    )
  ),
  xLabel = "Measurement A",
  yLabel = "Measurement B",
  colorScheme = "turbo"
)

mosaic(spec_realworld, realworld_data = realworld_data)

Use Cases

• Large Datasets: Essential for visualizing datasets with thousands or millions of points
• Density Visualization: Shows where data points are most concentrated
• Overplotting Solution: Replaces scatter plots when points overlap too much
• Pattern Detection: Reveals patterns that might be hidden in dense scatter plots
• Statistical Analysis: Useful for exploring bivariate distributions and correlations