A Chord Diagram With `ggplot`

Andy Grogan-Kaylor

In other earlier posts, I have looked at some ways to get an idea of what scales and demographic variables in my data are related to other scales and demographic variables in my data. One way of visualizing these relationships is a chord diagram.

In this post, I try creating a chord diagram using the fantastic new ggchord2 (Rennie, 2026a, 2026b).

Call Libraries

Show the code

library(faux) # simulated data

library(pander) # nice tables

library(skimr) # descriptive statistics

library(circlize) # circular graphs and chord diagrams

library(scales) # viridis color palettes

library(ggplot2) # beautiful graphs

library(ggchord2) # chord diagrams

Simulated Data

I again simulate some data.

Show the code

set.seed(54321) # random seed

# simulate data

cmat <- c(1.00, 0.30, 0.25, 0.15, 0.15, 0.15,
          0.30, 1.00, 0.00, 0.50, 0.10, 0.15,
          0.25, 0.00, 1.00, 0.50, 0.50, 0.15,
          0.15, 0.50, 0.50, 1.00, 0.00, 0.15,
          0.15, 0.10, 0.50, 0.00, 1.00, 0.15,
          0.15, 0.15, 0.15, 0.15, 0.15, 1.00)

mydata <- rnorm_multi(n = 100, 
                   mu = c(0, 20, 100, 100, 100, 100),
                   sd = c(1, 5, 5, 5, 5, 5),
                   r = cmat, # correlation matrix
                   varnames = c("scaleA", 
                                "scaleB", 
                                "scaleC", 
                                "scaleD", 
                                "scaleE", 
                                "scaleF"),
                   empirical = FALSE)

Look At The Data

Show the code

pander(head(mydata)) # nicely formatted table

scaleA	scaleB	scaleC	scaleD	scaleE	scaleF
1.602	23.34	99.84	97.63	102.1	101.5
1.452	28.85	101.6	99.86	105.9	99.57
0.5092	20.43	105.6	101.7	102.4	101
1.974	21.07	108.3	106.9	102	106.1
-0.9652	17.06	102	106.7	101.5	95.6
0.7906	22.49	104	111.2	96.67	98.47

Show the code

skim(mydata) # descriptive statistics

Variable type: numeric

skim_variable	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
scaleA	1	-0.05	1.09	-2.65	-0.80	-0.08	0.65	2.84	▂▆▇▅▁
scaleB	1	20.13	4.70	7.58	16.83	20.32	23.44	29.89	▁▅▇▆▃
scaleC	1	100.11	4.84	88.04	96.77	100.35	102.94	111.84	▂▃▇▃▂
scaleD	1	100.75	4.86	88.29	97.52	100.31	103.65	113.34	▁▆▇▆▁
scaleE	1	100.14	4.95	82.39	97.41	100.81	103.05	112.10	▁▂▅▇▁
scaleF	1	100.24	5.77	86.48	95.77	100.42	103.80	113.96	▁▇▇▅▂

Chord Diagram

For ggchord to work, I need data on the correlations of different scales, but need this correlation data to be formatted as a list of sources and destinations.

I therefore first create a correlation matrix of my data, and then reformat this correlation matrix as a data frame of sources, destinations, and correlations.

https://www.tutorialspoint.com/ has a page suggesting how to achieve this using Base R.

Show the code

cor_mat <- cor(mydata, # correlation matrix
               use = "complete.obs")

diag(cor_mat) <- 0 # set diagonal to 0

# thanks tutorialspoint.com!
mychorddata <- as.data.frame(as.table(cor_mat))

head(mychorddata) # look at the data

    Var1   Var2      Freq
1 scaleA scaleA 0.0000000
2 scaleB scaleA 0.5407251
3 scaleC scaleA 0.4047779
4 scaleD scaleA 0.2703827
5 scaleE scaleA 0.3092786
6 scaleF scaleA 0.1796353

Using ggchord is then relatively straightforward. I use geom_chord_diagram which adds all of the required geometries of a chord diagram, but could use other geoms to add each geometry separately.

Show the code

# pal_viridis()(5) # 5 viridis colors

ggplot(data = mychorddata, 
       mapping = aes(
         source = Var1,
         target = Var2,
         freq = Freq,
         fill = Var1)) + 
  geom_chord_diagram() + # add all the chord geoms
  # geom_chord_arcs() + # selecting specific 
  # geom_chord_sectors() + # chord geoms allows 
  # geom_chord_labels() + # for more customization
  scale_fill_viridis_d(name = "scale") + # nice colors
  labs(title = "Relationship of Scales In Our Data") +
  theme_void() +
  coord_fixed(clip = "off") # fixed coordinates; don't clip labels

References

Rennie, N. (2026a). ggchord2: Chord diagrams with ’ggplot2’. https://doi.org/10.32614/CRAN.package.ggchord2

Rennie, N. (2026b). Chord diagrams in R with ‘ggchord2‘. https://nrennie.rbind.io/blog/chord-diagrams-ggchord2/