-
Notifications
You must be signed in to change notification settings - Fork 16
/
README.Rmd
300 lines (231 loc) · 11.1 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
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r, echo = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
fig.align = "center",
out.width = "80%",
comment = "#>",
fig.path = "man/figures/README-",
echo = TRUE,
warning = FALSE,
message = FALSE
)
```
# graphlayouts <img src="man/figures/logo.png" align="right"/>
[](https://github.com/schochastics/graphlayouts/actions)
[](https://cran.r-project.org/package=graphlayouts)
[](https://CRAN.R-project.org/package=graphlayouts)
[](https://CRAN.R-project.org/package=graphlayouts)
[](https://app.codecov.io/gh/schochastics/graphlayouts?branch=main)
[](https://doi.org/10.5281/zenodo.7870213)
[](https://doi.org/10.21105/joss.05238)
This package implements some graph layout algorithms that are not available
in `igraph`.
**A detailed introductory tutorial for graphlayouts and ggraph can be found [here](https://schochastics.github.io/netVizR/).**
The package implements the following algorithms:
- Stress majorization ([Paper](https://graphviz.gitlab.io/_pages/Documentation/GKN04.pdf))
- Quadrilateral backbone layout ([Paper](https://kops.uni-konstanz.de/entities/publication/3403ae27-3e43-4a7c-9818-4484433fcce8))
- flexible radial layouts ([Paper](https://www.uni-konstanz.de/algo/publications/bp-mfrl-11.pdf))
- sparse stress ([Paper](https://arxiv.org/abs/1608.08909))
- pivot MDS ([Paper](https://kops.uni-konstanz.de/bitstream/handle/123456789/5741/bp_empmdsld_06.pdf?sequence=1&isAllowed=y))
- dynamic layout for longitudinal data ([Paper](https://kops.uni-konstanz.de/bitstream/handle/123456789/20924/Brandes_209246.pdf?sequence=2))
- spectral layouts (adjacency/Laplacian)
- a simple multilevel layout
- a layout algorithm using UMAP
- group based centrality and focus layouts which keeps groups of nodes close in the same range on the concentric circle
## Install
```{r install,eval=FALSE}
# dev version
remotes::install_github("schochastics/graphlayouts")
# CRAN
install.packages("graphlayouts")
```
## Stress Majorization: Connected Network
*This example is a bit of a special case since it exploits some weird issues in igraph.*
```{r example}
library(igraph)
library(ggraph)
library(graphlayouts)
set.seed(666)
pa <- sample_pa(1000, 1, 1, directed = F)
ggraph(pa, layout = "nicely") +
geom_edge_link0(width = 0.2, colour = "grey") +
geom_node_point(col = "black", size = 0.3) +
theme_graph()
ggraph(pa, layout = "stress") +
geom_edge_link0(width = 0.2, colour = "grey") +
geom_node_point(col = "black", size = 0.3) +
theme_graph()
```
## Stress Majorization: Unconnected Network
Stress majorization also works for networks with several components. It relies
on a bin packing algorithm to efficiently put the components in a rectangle, rather than a circle.
```{r example_un}
set.seed(666)
g <- disjoint_union(
sample_pa(10, directed = FALSE),
sample_pa(20, directed = FALSE),
sample_pa(30, directed = FALSE),
sample_pa(40, directed = FALSE),
sample_pa(50, directed = FALSE),
sample_pa(60, directed = FALSE),
sample_pa(80, directed = FALSE)
)
ggraph(g, layout = "nicely") +
geom_edge_link0() +
geom_node_point() +
theme_graph()
ggraph(g, layout = "stress", bbox = 40) +
geom_edge_link0() +
geom_node_point() +
theme_graph()
```
## Backbone Layout
Backbone layouts are helpful for drawing hairballs.
```{r hairball,eval = FALSE}
set.seed(665)
# create network with a group structure
g <- sample_islands(9, 40, 0.4, 15)
g <- simplify(g)
V(g)$grp <- as.character(rep(1:9, each = 40))
ggraph(g, layout = "stress") +
geom_edge_link0(colour = rgb(0, 0, 0, 0.5), width = 0.1) +
geom_node_point(aes(col = grp)) +
scale_color_brewer(palette = "Set1") +
theme_graph() +
theme(legend.position = "none")
```
<img src="man/figures/README-hairball-1.png" width="80%" style="display: block; margin: auto;" />
The backbone layout helps to uncover potential group structures based on edge
embeddedness and puts more emphasis on this structure in the layout.
```{r backbone,eval=FALSE}
bb <- layout_as_backbone(g, keep = 0.4)
E(g)$col <- FALSE
E(g)$col[bb$backbone] <- TRUE
ggraph(g, layout = "manual", x = bb$xy[, 1], y = bb$xy[, 2]) +
geom_edge_link0(aes(col = col), width = 0.1) +
geom_node_point(aes(col = grp)) +
scale_color_brewer(palette = "Set1") +
scale_edge_color_manual(values = c(rgb(0, 0, 0, 0.3), rgb(0, 0, 0, 1))) +
theme_graph() +
theme(legend.position = "none")
```
<img src="man/figures/README-backbone-1.png" width="80%" style="display: block; margin: auto;" />
## Radial Layout with Focal Node
The function `layout_with_focus()` creates a radial layout around a focal node.
All nodes with the same distance from the focal node are on the same circle.
```{r flex_focus,eval=FALSE}
library(igraphdata)
library(patchwork)
data("karate")
p1 <- ggraph(karate, layout = "focus", focus = 1) +
draw_circle(use = "focus", max.circle = 3) +
geom_edge_link0(edge_color = "black", edge_width = 0.3) +
geom_node_point(aes(fill = as.factor(Faction)), size = 2, shape = 21) +
scale_fill_manual(values = c("#8B2323", "#EEAD0E")) +
theme_graph() +
theme(legend.position = "none") +
coord_fixed() +
labs(title = "Focus on Mr. Hi")
p2 <- ggraph(karate, layout = "focus", focus = 34) +
draw_circle(use = "focus", max.circle = 4) +
geom_edge_link0(edge_color = "black", edge_width = 0.3) +
geom_node_point(aes(fill = as.factor(Faction)), size = 2, shape = 21) +
scale_fill_manual(values = c("#8B2323", "#EEAD0E")) +
theme_graph() +
theme(legend.position = "none") +
coord_fixed() +
labs(title = "Focus on John A.")
p1 + p2
```
<img src="man/figures/README-flex_focus-1.png" width="80%" style="display: block; margin: auto;" />
## Radial Centrality Layout
The function `layout_with_centrality` creates a radial layout around the node with the
highest centrality value. The further outside a node is, the more peripheral it is.
```{r flex_cent}
library(igraphdata)
library(patchwork)
data("karate")
bc <- betweenness(karate)
p1 <- ggraph(karate, layout = "centrality", centrality = bc, tseq = seq(0, 1, 0.15)) +
draw_circle(use = "cent") +
annotate_circle(bc, format = "", pos = "bottom") +
geom_edge_link0(edge_color = "black", edge_width = 0.3) +
geom_node_point(aes(fill = as.factor(Faction)), size = 2, shape = 21) +
scale_fill_manual(values = c("#8B2323", "#EEAD0E")) +
theme_graph() +
theme(legend.position = "none") +
coord_fixed() +
labs(title = "betweenness centrality")
cc <- closeness(karate)
p2 <- ggraph(karate, layout = "centrality", centrality = cc, tseq = seq(0, 1, 0.2)) +
draw_circle(use = "cent") +
annotate_circle(cc, format = "scientific", pos = "bottom") +
geom_edge_link0(edge_color = "black", edge_width = 0.3) +
geom_node_point(aes(fill = as.factor(Faction)), size = 2, shape = 21) +
scale_fill_manual(values = c("#8B2323", "#EEAD0E")) +
theme_graph() +
theme(legend.position = "none") +
coord_fixed() +
labs(title = "closeness centrality")
p1 + p2
```
## Large graphs
`graphlayouts` implements two algorithms for visualizing large networks (<100k nodes).
`layout_with_pmds()` is similar to `layout_with_mds()` but performs the multidimensional scaling
only with a small number of pivot nodes. Usually, 50-100 are enough to obtain similar results to the
full MDS.
`layout_with_sparse_stress()` performs stress majorization only with a small number of pivots (~50-100).
The runtime performance is inferior to pivotMDS but the quality is far superior.
A comparison of runtimes and layout quality can be found in the [wiki](https://github.com/schochastics/graphlayouts/wiki/)
**tl;dr**: both layout algorithms appear to be faster than the fastest igraph algorithm `layout_with_drl()`.
Below are two examples of layouts generated for large graphs using `layout_with_sparse_stress()`
<img src="man/figures/rt-net.png" width="80%" style="display: block; margin: auto;" />
A retweet network with 18k nodes and 61k edges
<img src="man/figures/squad_network2022_small.png" width="80%" style="display: block; margin: auto;" />
A network of football players with 165K nodes and 6M edges.
## dynamic layouts
`layout_as_dynamic()` allows you to visualize snapshots of longitudinal network data. Nodes are anchored
with a reference layout and only moved slightly in each wave depending on deleted/added edges.
In this way, it is easy to track down specific nodes throughout time. Use `patchwork` to put the
individual plots next to each other.
```{r dynamic,eval = FALSE}
# remotes::install_github("schochastics/networkdata")
library(networkdata)
# longitudinal dataset of friendships in a school class
data("s50")
xy <- layout_as_dynamic(s50, alpha = 0.2)
pList <- vector("list", length(s50))
for (i in seq_along(s50)) {
pList[[i]] <- ggraph(s50[[i]], layout = "manual", x = xy[[i]][, 1], y = xy[[i]][, 2]) +
geom_edge_link0(edge_width = 0.6, edge_colour = "grey66") +
geom_node_point(shape = 21, aes(fill = as.factor(smoke)), size = 3) +
geom_node_text(aes(label = 1:50), repel = T) +
scale_fill_manual(
values = c("forestgreen", "grey25", "firebrick"),
labels = c("no", "occasional", "regular"),
name = "smoking",
guide = ifelse(i != 2, "none", "legend")
) +
theme_graph() +
theme(legend.position = "bottom") +
labs(title = paste0("Wave ", i))
}
wrap_plots(pList)
```
<img src="man/figures/dynamic_ex.png" width="80%" style="display: block; margin: auto;" />
## Layout manipulation
The functions `layout_mirror()` and `layout_rotate()` can be used to manipulate an existing layout
<img src="man/figures/layout_manipulation.png" width="80%" style="display: block; margin: auto;" />
# How to reach out?
### Where do I report bugs?
Simply [open an issue](https://github.com/schochastics/graphlayouts/issues/new) on GitHub.
### How do I contribute to the package?
If you have an idea (but no code yet), [open an issue](https://github.com/schochastics/graphlayouts/issues/new) on GitHub. If you want to contribute with a specific feature and have the code ready, fork the repository, add your code, and create a pull request.
### Do you need support?
The easiest way is to [open an issue](https://github.com/schochastics/graphlayouts/issues/new) - this way, your question is also visible to others who may face similar problems.
### Code of Conduct
Please note that the graphlayouts project is released with a [Contributor Code of Conduct](https://contributor-covenant.org/version/2/1/CODE_OF_CONDUCT.html). By contributing to this project, you agree to abide by its terms.