Workflow: Data Scientist Creating Accessible Dashboards • huerd

The Scenario

You’re a data scientist creating a dashboard to visualize patient outcomes across treatment groups. Your requirements are:

6 distinct colors for different treatment groups
Colorblind-safe for accessibility compliance
Include company brand colors (a specific blue and orange)
Easy integration with ggplot2

This vignette walks through how huerd makes this workflow simple and reliable.

Quick Start: The Simple Path

For many use cases, quick_palette() is all you need:

library(huerd)

# Generate a 6-color accessible palette
palette <- quick_palette(6)
print(palette)
#> 
#> -- huerd Color Palette (6 colors) --
#> Colors:
#> [ 1] #4C0E00
#> [ 2] #002E8E
#> [ 3] #9A00FF
#> [ 4] #FF0025
#> [ 5] #00FFF5
#> [ 6] #EFE600
#> 
#> -- Quality Metrics Summary --
#> * Min. Perceptual Distance (OKLAB): 0.233
#> * Optimizer Performance Ratio      : 63.7%
#> * Min. CVD-Safe Distance (OKLAB)  : 0.201
#> 
#> -- Generation Details --
#> * Optimizer Iterations: 401
#> * Optimizer Status: NLOPT_XTOL_REACHED: Optimization stopped because xtol_rel or xtol_abs (above) was reached.

Want to include your brand colors? Just add them:

brand_palette <- quick_palette(
  n = 6,
  brand_colors = c("#1f77b4", "#ff7f0e")  # Your blue and orange
)
print(brand_palette)
#> 
#> -- huerd Color Palette (6 colors) --
#> Colors:
#> [ 1] #430096
#> [ 2] #1F77B4
#> [ 3] #FF7F0E
#> [ 4] #B19EFF
#> [ 5] #00FFFF
#> [ 6] #FFEE00
#> 
#> -- Quality Metrics Summary --
#> * Min. Perceptual Distance (OKLAB): 0.230
#> * Optimizer Performance Ratio      : 63.0%
#> * Min. CVD-Safe Distance (OKLAB)  : 0.173
#> 
#> -- Generation Details --
#> * Optimizer Iterations: 210
#> * Optimizer Status: NLOPT_XTOL_REACHED: Optimization stopped because xtol_rel or xtol_abs (above) was reached.

Using Palettes with ggplot2

huerd provides native ggplot2 integration. No more scale_color_manual():

library(ggplot2)

# Simulate treatment outcome data
set.seed(123)
treatment_data <- data.frame(
  patient_id = 1:180,
  treatment = rep(c("Control", "Drug A", "Drug B", "Drug C", "Combo 1", "Combo 2"), each = 30),
  outcome = c(
    rnorm(30, 50, 10),
    rnorm(30, 55, 10),
    rnorm(30, 60, 10),
    rnorm(30, 58, 10),
    rnorm(30, 65, 10),
    rnorm(30, 62, 10)
  ),
  time_weeks = rep(1:30, 6)
)

# Plot with automatic huerd palette
ggplot(treatment_data, aes(x = time_weeks, y = outcome, color = treatment)) +
  geom_smooth(method = "loess", se = FALSE) +
  scale_color_huerd() +
  labs(
    title = "Patient Outcomes by Treatment Group",
    x = "Weeks",
    y = "Outcome Score",
    color = "Treatment"
  ) +
  theme_minimal()
#> `geom_smooth()` using formula = 'y ~ x'

With Brand Colors

Include your brand colors directly in the scale:

ggplot(treatment_data, aes(x = time_weeks, y = outcome, color = treatment)) +
  geom_smooth(method = "loess", se = FALSE) +
  scale_color_huerd(brand_colors = c("#1f77b4", "#ff7f0e")) +
  labs(
    title = "Patient Outcomes by Treatment Group",
    subtitle = "Using brand colors",
    x = "Weeks",
    y = "Outcome Score",
    color = "Treatment"
  ) +
  theme_minimal()
#> `geom_smooth()` using formula = 'y ~ x'

Using a Pre-Generated Palette

For consistent colors across multiple plots, generate once and reuse:

# Generate and save your palette
my_palette <- generate_palette(
  n = 6,
  include_colors = c("#1f77b4", "#ff7f0e"),
  progress = FALSE
)

# Use in multiple plots
p1 <- ggplot(treatment_data, aes(x = treatment, y = outcome, fill = treatment)) +
  geom_boxplot() +
  scale_fill_huerd(palette = my_palette) +
  labs(title = "Distribution by Treatment") +
  theme_minimal() +
  theme(legend.position = "none")

p2 <- ggplot(treatment_data, aes(x = outcome, fill = treatment)) +
  geom_density(alpha = 0.6) +
  scale_fill_huerd(palette = my_palette) +
  labs(title = "Outcome Density") +
  theme_minimal()

# Display side by side
gridExtra::grid.arrange(p1, p2, ncol = 2)

Verifying Accessibility

Before finalizing your dashboard, verify accessibility:

Quick Check

is_cvd_safe(my_palette)
#> [1] TRUE

Detailed Assessment

interpret_palette_quality(my_palette)
#> 
#> ── Palette Quality Assessment ──
#> 
#> This 6-color palette is well optimized (42% of theoretical maximum). Excellent
#> - colors are highly distinct and easy to differentiate
#> 
#> ── Distinctness
#> Excellent - colors are highly distinct and easy to differentiate
#> 
#> ── Accessibility
#> Good - palette should work for most viewers with CVD

Full Evaluation

evaluation <- evaluate_palette(my_palette)

# Key metrics for your accessibility report
cat("Minimum color distance:", round(evaluation$distances$min, 3), "\n")
#> Minimum color distance: 0.152
cat("CVD worst-case distance:", round(evaluation$cvd_safety$worst_case_min_distance, 3), "\n")
#> CVD worst-case distance: 0.098
cat("Performance ratio:", round(evaluation$distances$performance_ratio * 100, 1), "%\n")
#> Performance ratio: 41.7 %

Visual Diagnostics

For presentations or documentation, use the analysis dashboard:

plot_palette_analysis(my_palette)

Or the simple swatch view:

plot(my_palette)

Summary

For data scientists, huerd provides:

quick_palette() - Fast, sensible defaults
scale_color_huerd() / scale_fill_huerd() - Native ggplot2 integration
is_cvd_safe() - Quick accessibility verification
interpret_palette_quality() - Human-readable quality assessment
plot_palette_analysis() - Comprehensive visual diagnostics

This workflow ensures your visualizations are both beautiful and accessible.