Updates to better support dendrograms by jrowen · Pull Request #818 · plotly/plotly.R

jrowen · 2016-11-30T14:02:35Z

These changes were made to better support dendrograms, particularly those created with dendextend. It adds support for

fontface in geom_text
a rough translation of hjust and vjust in geom_text
-Inf and Inf in geom_rect
guide = "none"

Feedback is very welcome. Below is some sample code.

library(ggplot2)
library(plotly)
library(dendextend)

dend <- iris[1:30,-5] %>% dist %>% hclust %>% as.dendrogram %>%
  set("branches_k_color", k=3) %>% set("branches_lwd", c(1.5,1,1.5)) %>%
  set("branches_lty", c(1,1,3,1,1,2)) %>%
  set("labels_colors") %>% set("labels_cex", c(.9,1.2)) %>% 
  set("nodes_pch", 19) %>% set("nodes_col", c("orange", "black", "plum", NA))
ggd1 <- as.ggdend(dend)
p = ggplot(ggd1, horiz = TRUE, theme = NULL)
ggplotly(p)

p1 = ggplot(cars) +
  geom_point(aes(x = speed, y = dist)) +
  geom_rect(xmin = 5, xmax = 10, ymin = -Inf, ymax = Inf)
ggplotly(p1)

p2 = ggplot(cars) +
  geom_point(aes(x = speed, y = dist)) +
  geom_rect(ymin = 25, ymax = 50, xmin = -Inf, xmax = Inf)
ggplotly(p2)

p3 = ggplot(cars) +
  geom_point(aes(x = speed, y = dist)) +
  geom_rect(xmin = 5, xmax = 10, ymin = -Inf, ymax = Inf) +
  coord_flip()
ggplotly(p3)

p4 = ggplot(cars) +
  geom_point(aes(x = speed, y = dist)) +
  geom_rect(ymin = 25, ymax = 50, xmin = -Inf, xmax = Inf) +
  coord_flip()
ggplotly(p4)

p5 = ggplot(data.frame(x = seq(5, 25, 5), y = 60)) +
  geom_point(aes(x = x, y = y)) +
  geom_text(x = 5, y = 60, label = "nothing") +
  geom_text(x = 10, y = 60, label = "bold", fontface = "bold", hjust = 0, vjust = 0) +
  geom_text(x = 15, y = 60, label = "italic", fontface = "italic", hjust = 1, vjust = 1) +
  geom_text(x = 20, y = 60, label = "bold italic", fontface = "bold italic", hjust = 0, vjust = 1) +
  geom_text(x = 25, y = 60, label = "plain", fontface = "plain", hjust = 1, vjust = 0)
ggplotly(p5)

Conflicts: R/layers2traces.R

cpsievert · 2016-11-30T16:16:13Z

R/layers2traces.R

+    grepl("italic", data[["fontface"]]),
+    paste0("<i>", text, "</i>"),
+    text
+  )


I'd prefer if this was done with if() instead of ifelse()

I was using ifelse as I was anticipating data[["fontface"]] could be a vector (I think it can be specified via aes). Is this an incorrect assumption?

oops, yea, but in that case I prefer something like:

text[grepl("italic", data[["fontface"]])] <- paste0("<i>", text, "</i>")

cpsievert · 2016-11-30T16:17:34Z

R/layers2traces.R

+            cbind(x = ifelse(xmax == Inf, layout$x_max, xmax),
+                  y = ifelse(ymin == -Inf, layout$y_min, ymin), others))
+    })
+  }


This is a lot of redundant code...

Good point, I've simplified things a bit.

cpsievert · 2016-11-30T16:19:28Z

R/ggplotly.R


  # if theme(legend.position = "none") is used, don't show a legend _or_ guide
+  npscales$scales <- Filter(function(x) x$guide != "none", npscales$scales)
  if (npscales$n() == 0 || identical(theme$legend.position, "none")) {


I see this as potentially dangerous -- a scale without a guide doesn't necessarily mean it doesn't exist.

I would prefer to instead set layout.showlegend to FALSE for discrete scales and remove the colorbar for continuous ones.

How about instead handling the discrete scale by updating this line to something like

sc$is_discrete() && sc$guide != "none"

That won't work either for similar reasons

I made one more attempt, hopefully we're close. Thanks.

cpsievert · 2017-01-02T20:48:14Z

Sorry for the delay @jrowen, this is looking pretty good...before I merge, would you mind writing a couple simple tests?

jrowen · 2017-02-20T22:15:11Z

Sorry, I too have been away from this for a while. I'll see if I can pull those together shortly. Thanks.

talgalili · 2017-04-12T09:20:27Z

Hi @jrowen
Any chance you could get some tests made for this?
This would be helpful for heatmaply (since it now has a plot_method="plotly" which could use these features).

Thanks!

jrowen · 2017-04-12T13:17:41Z

I added some checks for geom_text and geom_rect, but I'm not sure how to test the guide = "none" change without introducing a dependency on dendextend (see code below which now works as expected). It also looks like the R:devel build is failing with the R:release build is fine.

library(ggplot2)
library(plotly)
library(dendextend)

dend <- iris[1:30,-5] %>% dist %>% hclust %>% as.dendrogram %>%
  set("branches_k_color", k=3) %>% set("branches_lwd", c(1.5,1,1.5)) %>%
  set("branches_lty", c(1,1,3,1,1,2)) %>%
  set("labels_colors") %>% set("labels_cex", c(.9,1.2)) %>% 
  set("nodes_pch", 19) %>% set("nodes_col", c("orange", "black", "plum", NA))
ggd1 <- as.ggdend(dend)
p = ggplot(ggd1, horiz = TRUE, theme = NULL)
ggplotly(p)

cpsievert · 2017-04-13T21:54:20Z

Don't worry about the build failure @jrowen. I'll have a closer look here next week

talgalili · 2017-05-10T08:37:12Z

Hi Carson,
Any chance you can give this a look?

Thanks,
Tal

cpsievert · 2017-05-11T00:46:23Z

I may get to this next week. #929 will take precedence.

cpsievert · 2018-05-29T17:22:58Z

Hi @jrowen, sorry for forgetting about this PR! I thought I was going to have the bandwidth to complete #929, but I'm not sure anymore...I have a couple small questions about this PR, but we should try to get this merged before I submit to CRAN by the end of June

cpsievert · 2018-05-29T17:24:12Z

R/layers2traces.R

+    textposition = paste0(
+      ifelse(data[["vjust"]] < 0.5, "top ",
+             ifelse(data[["vjust"]] > 0.5, "bottom ", "")
+      ),


Shouldn't the trailing whitespace be removed here (e.g., "top" not "top ")?

Ahh, nevermind. I see why it's done this why

I think this should default to "middle" (not "") though

cpsievert · 2018-05-29T17:29:13Z

R/layers2traces.R

+                y = ifelse(ymax == Inf, y_max, ymax), others),
+          cbind(x = ifelse(xmax == Inf, x_max, xmax),
+                y = ifelse(ymin == -Inf, y_min, ymin), others))
  })


Perhaps it's better (more efficient) to use is.finite() here?

cpsievert · 2018-05-29T17:36:03Z

tests/testthat/test-ggplot-rect.R

  expect_match(info$data[[1]]$fillcolor, "rgba\\(0,0,0,0\\.0[6]+")
 })

+p1 = ggplot(data.frame(x = 1, y = 1)) +


It'd be great to have another test that specifying ymin = 0.5/ymax = 1.5 gives the same result

talgalili · 2021-01-29T10:54:41Z

Hey @cpsievert :)
Any chance to add this support?

jrowen added 8 commits November 30, 2016 01:29

Improvements for dendrograms

a6663c4

Improvements for dendrograms

208eed4

Merge branch 'master' of https://github.com/jrowen/plotly

9489610

Conflicts: R/layers2traces.R

Fix rect to handle coord flip

4dcdebb

Fix fontface

10936b4

Fix fontface

f7cce29

Fixed rect

7862f00

Cleanup hjust

3e63f3c

jrowen mentioned this pull request Nov 30, 2016

ggplotly conversion of dendextend ggplot #807

Closed

cpsievert reviewed Nov 30, 2016

View reviewed changes

jrowen added 4 commits November 30, 2016 16:33

Clean-up GeomRect

16d097b

Update fontface approach

dac04d2

Updated approach for guide = none

180fcfd

Another guide = none attempt

183b39f

Added geom_text and geom_rect tests

88110c0

cpsievert added this to the ggplotly parity milestone May 11, 2017

cpsievert reviewed May 29, 2018

View reviewed changes

Conversation

jrowen commented Nov 30, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cpsievert commented Jan 2, 2017

Uh oh!

jrowen commented Feb 20, 2017

Uh oh!

talgalili commented Apr 12, 2017

Uh oh!

jrowen commented Apr 12, 2017

Uh oh!

cpsievert commented Apr 13, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

talgalili commented May 10, 2017

Uh oh!

cpsievert commented May 11, 2017

Uh oh!

cpsievert commented May 29, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

cpsievert May 29, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

talgalili commented Jan 29, 2021

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

cpsievert commented Apr 13, 2017 •

edited

Loading

cpsievert commented May 29, 2018 •

edited

Loading

cpsievert May 29, 2018 •

edited

Loading