-
Notifications
You must be signed in to change notification settings - Fork 10
/
Copy pathREADME.Rmd
384 lines (321 loc) · 12.8 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
library(edgebundle)
```
# edgebundle <img src="man/figures/logo.png" align="right"/>
<!-- badges: start -->
[](https://github.com/schochastics/edgebundle/actions/workflows/R-CMD-check.yaml)
[](https://CRAN.R-project.org/package=edgebundle)
[](https://CRAN.R-project.org/package=edgebundle)
<!-- badges: end -->
An R package that implements several edge bundling/flow and metro map algorithms. So far it includes
- Force directed edge bundling
- Stub bundling ([paper](https://www.uni-konstanz.de/mmsp/pubsys/publishedFiles/NoBr13.pdf))
- Hammer bundling ([python code](https://datashader.org/_modules/datashader/bundling.html))
- Edge-path bundling ([paper](https://arxiv.org/abs/2108.05467))
- TNSS flow map ([paper](https://doi.org/10.1080/15230406.2018.1437359))
- Multicriteria Metro map layout ([paper](https://doi.org/10.1109/TVCG.2010.24))
**[ggraph 2.2.0](https://www.data-imaginist.com/posts/2024-02-15-ggraph-2-2-0/) supports edge bundling natively via `geom_edge_bundle_*()` functions. This means that parts of this package are now deprecated.**
## Installation
The package is available on CRAN.
```{r cran, eval=FALSE}
install.packages("edgebundle")
```
The developer version can be installed with
``` {r dev, eval=FALSE}
# install.packages("remotes")
remotes::install_github("schochastics/edgebundle")
```
Note that `edgebundle` imports `reticulate` and uses a pretty big python library (datashader).
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>"
)
```
```{r setup,warning=FALSE,message=FALSE}
library(edgebundle)
library(igraph)
```
## Edge bundling
The expected input of each edge bundling function is a graph (igraph/network or tbl_graph object) and a node layout.
All functions return a data frame of points along the edges of the network that can be plotted with {{ggplot2}} using `geom_path()` or
`geom_bezier()` for `edge_bundle_stub()`.
```{r example,message=FALSE}
library(igraph)
g <- graph_from_edgelist(
matrix(c(1, 12, 2, 11, 3, 10, 4, 9, 5, 8, 6, 7), ncol = 2, byrow = T), F
)
xy <- cbind(c(rep(0, 6), rep(1, 6)), c(1:6, 1:6))
fbundle <- edge_bundle_force(g, xy, compatibility_threshold = 0.1)
head(fbundle)
```
The result can be visualized as follows.
```{r plot,message=FALSE,fig.align='center'}
library(ggplot2)
ggplot(fbundle) +
geom_path(aes(x, y, group = group, col = as.factor(group)),
size = 2, show.legend = FALSE
) +
geom_point(data = as.data.frame(xy), aes(V1, V2), size = 5) +
theme_void()
# simple edge-path bundling example
g <- graph_from_edgelist(matrix(c(1, 2, 1, 6, 1, 4, 2, 3, 3, 4, 4, 5, 5, 6),
ncol = 2, byrow = TRUE
), FALSE)
xy <- cbind(c(0, 10, 25, 40, 50, 50), c(0, 15, 25, 15, 0, -10))
res <- edge_bundle_path(g, xy, max_distortion = 2, weight_fac = 2, segments = 50)
ggplot() +
geom_path(
data = res, aes(x, y, group = group, col = as.factor(group)),
size = 2, show.legend = FALSE
) +
geom_point(data = as.data.frame(xy), aes(V1, V2), size = 5) +
scale_color_manual(values = c("grey66", "firebrick3", "firebrick3", rep("grey66", 4))) +
theme_void()
```
For `edge_bundle_stub()`, you need `geom_bezier()` from the package {{ggforce}}.
```{r bezier,message=FALSE,fig.align='center'}
library(ggforce)
g <- graph.star(10, "undirected")
xy <- matrix(c(
0, 0,
cos(90 * pi / 180), sin(90 * pi / 180),
cos(80 * pi / 180), sin(80 * pi / 180),
cos(70 * pi / 180), sin(70 * pi / 180),
cos(330 * pi / 180), sin(330 * pi / 180),
cos(320 * pi / 180), sin(320 * pi / 180),
cos(310 * pi / 180), sin(310 * pi / 180),
cos(210 * pi / 180), sin(210 * pi / 180),
cos(200 * pi / 180), sin(200 * pi / 180),
cos(190 * pi / 180), sin(190 * pi / 180)
), ncol = 2, byrow = TRUE)
sbundle <- edge_bundle_stub(g, xy, beta = 90)
ggplot(sbundle) +
geom_bezier(aes(x, y, group = group), size = 1.5, col = "grey66") +
geom_point(data = as.data.frame(xy), aes(V1, V2), size = 5) +
theme_void()
```
The typical edge bundling benchmark uses a dataset on us flights, which is included in the package.
```{r example-code,eval = FALSE}
g <- us_flights
xy <- cbind(V(g)$longitude, V(g)$latitude)
verts <- data.frame(x = V(g)$longitude, y = V(g)$latitude)
fbundle <- edge_bundle_force(g, xy, compatibility_threshold = 0.6)
sbundle <- edge_bundle_stub(g, xy)
hbundle <- edge_bundle_hammer(g, xy, bw = 0.7, decay = 0.5)
pbundle <- edge_bundle_path(g, xy, max_distortion = 12, weight_fac = 2, segments = 50)
states <- map_data("state")
p1 <- ggplot() +
geom_polygon(
data = states, aes(long, lat, group = group),
col = "white", size = 0.1, fill = NA
) +
geom_path(
data = fbundle, aes(x, y, group = group),
col = "#9d0191", size = 0.05
) +
geom_path(
data = fbundle, aes(x, y, group = group),
col = "white", size = 0.005
) +
geom_point(
data = verts, aes(x, y),
col = "#9d0191", size = 0.25
) +
geom_point(
data = verts, aes(x, y),
col = "white", size = 0.25, alpha = 0.5
) +
geom_point(
data = verts[verts$name != "", ],
aes(x, y), col = "white", size = 3, alpha = 1
) +
labs(title = "Force Directed Edge Bundling") +
ggraph::theme_graph(background = "black") +
theme(plot.title = element_text(color = "white"))
p2 <- ggplot() +
geom_polygon(
data = states, aes(long, lat, group = group),
col = "white", size = 0.1, fill = NA
) +
geom_path(
data = hbundle, aes(x, y, group = group),
col = "#9d0191", size = 0.05
) +
geom_path(
data = hbundle, aes(x, y, group = group),
col = "white", size = 0.005
) +
geom_point(
data = verts, aes(x, y),
col = "#9d0191", size = 0.25
) +
geom_point(
data = verts, aes(x, y),
col = "white", size = 0.25, alpha = 0.5
) +
geom_point(
data = verts[verts$name != "", ], aes(x, y),
col = "white", size = 3, alpha = 1
) +
labs(title = "Hammer Edge Bundling") +
ggraph::theme_graph(background = "black") +
theme(plot.title = element_text(color = "white"))
alpha_fct <- function(x, b = 0.01, p = 5, n = 20) {
(1 - b) * (2 / (n - 1))^p * abs(x - (n - 1) / 2)^p + b
}
p3 <- ggplot() +
geom_polygon(
data = states, aes(long, lat, group = group),
col = "white", size = 0.1, fill = NA
) +
ggforce::geom_bezier(
data = sbundle, aes(x, y,
group = group,
alpha = alpha_fct(..index.. * 20)
), n = 20,
col = "#9d0191", size = 0.1, show.legend = FALSE
) +
ggforce::geom_bezier(
data = sbundle, aes(x, y,
group = group,
alpha = alpha_fct(..index.. * 20)
), n = 20,
col = "white", size = 0.01, show.legend = FALSE
) +
geom_point(
data = verts, aes(x, y),
col = "#9d0191", size = 0.25
) +
geom_point(
data = verts, aes(x, y),
col = "white", size = 0.25, alpha = 0.5
) +
geom_point(
data = verts[verts$name != "", ], aes(x, y),
col = "white", size = 3, alpha = 1
) +
labs(title = "Stub Edge Bundling") +
ggraph::theme_graph(background = "black") +
theme(plot.title = element_text(color = "white"))
p4 <- ggplot() +
geom_polygon(
data = states, aes(long, lat, group = group),
col = "white", size = 0.1, fill = NA
) +
geom_path(
data = pbundle, aes(x, y, group = group),
col = "#9d0191", size = 0.05
) +
geom_path(
data = pbundle, aes(x, y, group = group),
col = "white", size = 0.005
) +
geom_point(
data = verts, aes(x, y),
col = "#9d0191", size = 0.25
) +
geom_point(
data = verts, aes(x, y),
col = "white", size = 0.25, alpha = 0.5
) +
geom_point(
data = verts[verts$name != "", ], aes(x, y),
col = "white", size = 3, alpha = 1
) +
labs(title = "Edge-Path Bundling") +
ggraph::theme_graph(background = "black") +
theme(plot.title = element_text(color = "white"))
p1
p2
p3
p4
```
```{r,fig.width=8,fig.height=6,fig.align='center',out.width="95%",echo=FALSE}
knitr::include_graphics("man/figures/flights_fdeb.png")
knitr::include_graphics("man/figures/flights_heb.png")
knitr::include_graphics("man/figures/flights_seb.png")
knitr::include_graphics("man/figures/flights_peb.png")
```
## Flow maps
A flow map is a type of thematic map that represent movements. It may thus be considered a hybrid of a map and a flow diagram.
The package so far implements a spatial one-to-many flow layout algorithm using
triangulation and approximate Steiner trees.
The function `tnss_tree()` expects a one-to-many flow network (i.e. a weighted star graph), a layout for the nodes and
a set of dummy nodes created with `tnss_dummies()`.
```{r example_flow,eval=FALSE}
library(ggraph)
xy <- cbind(state.center$x, state.center$y)[!state.name %in% c("Alaska", "Hawaii"), ]
xy_dummy <- tnss_dummies(xy, 4)
gtree <- tnss_tree(cali2010, xy, xy_dummy, 4, gamma = 0.9)
ggraph(gtree, "manual", x = V(gtree)$x, y = V(gtree)$y) +
geom_polygon(data = us, aes(long, lat, group = group), fill = "#FDF8C7", col = "black") +
geom_edge_link(aes(width = flow, col = sqrt((xy[root, 1] - ..x..)^2 + (xy[root, 2] - ..y..)^2)),
lineend = "round", show.legend = FALSE
) +
scale_edge_width(range = c(0.5, 4), trans = "sqrt") +
scale_edge_color_gradient(low = "#cc0000", high = "#0000cc") +
geom_node_point(aes(filter = tnss == "leaf"), size = 1) +
geom_node_point(aes(filter = (name == "California")), size = 5, shape = 22, fill = "#cc0000") +
theme_graph() +
labs(title = "Migration from California (2010) - Flow map")
```
```{r,fig.width=8,fig.height=6,fig.align='center',out.width="95%",echo=FALSE}
knitr::include_graphics("man/figures/cali2010_flow.png")
```
To smooth the tree, use `tnss_smooth()`. Note that this changes the object type and
you need to visualize it with {{ggplot2}} rather than {{ggraph}}.
```{r example_flow_smooth,eval=FALSE}
smooth_df <- tnss_smooth(gtree, bw = 5, n = 20)
ggplot() +
geom_polygon(data = us, aes(long, lat, group = group), fill = "#FDF8C7", col = "black") +
geom_path(
data = smooth_df, aes(x, y, group = destination, size = flow),
lineend = "round", col = "firebrick3", alpha = 1
) +
theme_void() +
scale_size(range = c(0.5, 3), guide = "none") +
labs(title = "Migration from California (2010) - Flow map smoothed")
```
```{r,fig.width=8,fig.height=6,fig.align='center',out.width="95%",echo=FALSE}
knitr::include_graphics("man/figures/cali2010_flow_smoothed.png")
```
See [this gallery](http://minard.schochastics.net/) for more examples and code.
## Metro Maps
Metro map(-like) graph drawing follow certain rules, such as octilinear edges. The algorithm
implemented in the packages uses hill-climbing to optimize several features desired in a metro map.
The package includes the metro map of Berlin as an example.
```{r metro_example, eval = FALSE}
# the algorithm has problems with parallel edges
g <- simplify(metro_berlin)
xy <- cbind(V(g)$lon, V(g)$lat) * 100
# the algorithm is not very stable. try playing with the parameters
xy_new <- metro_multicriteria(g, xy, l = 2, gr = 0.5, w = c(100, 100, 1, 1, 100), bsize = 35)
# geographic layout
ggraph(metro_berlin, "manual", x = xy[, 1], y = xy[, 2]) +
geom_edge_link0(aes(col = route_I_counts), edge_width = 2, show.legend = FALSE) +
geom_node_point(shape = 21, col = "white", fill = "black", size = 3, stroke = 0.5)
# schematic layout
ggraph(metro_berlin, "manual", x = xy_new[, 1], y = xy_new[, 2]) +
geom_edge_link0(aes(col = route_I_counts), edge_width = 2, show.legend = FALSE) +
geom_node_point(shape = 21, col = "white", fill = "black", size = 3, stroke = 0.5) +
theme_graph() +
labs(title = "Subway Network Berlin")
```
```{r,fig.width=8,fig.height=6,fig.align='center',out.width="95%",echo=FALSE}
knitr::include_graphics("man/figures/metro_berlin.png")
```
## Disclaimer
Edge bundling is able to produce neat looking network visualizations. However, they do not
necessarily enhance readability. After experimenting with several methods, it became quite
evident that the algorithms are very sensitive to the parameter settings (and often really only work in the showcase examples...). Consult the original literature (if they even provide any guidelines) or experiment yourself and **do not expect any miracles**.