A Chord Diagram

Andy Grogan-Kaylor

In previous posts, I 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.

In this post, I try using a chord diagram using chordDiagram from the circlize library (Gu et al., 2014).

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

Simulated Data

I 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

Show the code

# pal_viridis()(5) # 5 viridis colors

grid.col = c(scaleA = "#440154FF", 
             scaleB = "#3B528BFF", 
             scaleC = "#21908CFF", 
             scaleD = "#5DC863FF", 
             scaleE = "#FDE725FF")

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

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

chordDiagram(cor_mat, 
             annotationTrack = c("name", "grid"), 
             big.gap = 25,
             scale = TRUE, # scale equivalently
             grid.col = grid.col)

References

Gu, Z., Gu, L., Eils, R., Schlesner, M., & Brors, B. (2014). Circlize implements and enhances circular visualization in R. Bioinformatics, 30, 2811–2812.