Plotbuilder adaptation for jamovi – necessary or possible?

[yurismol]

Foe example I get the code as

ggplot(data, aes(x, y, label = rownames(data)))

and want to add another text, or legend

p + geom_text()

If this works, then it will be very useful.

For example in my code something like this work fine:

Code: Select all

p <- ggmice::plot_pattern(data=dat, square=TRUE, rotate=TRUE)
p <- p + ggplot2::scale_fill_manual(values=fill, labels=c(.("missing"), .("observed"))) +
            ggplot2::theme(text=ggplot2::element_text(size=ggtheme[[1]]$text$size))

[vinschger]

I appreciate your responses. I am aware that the existing Jamovi modules are already powerful; I am especially impressed by the usefulness of the modules from @vjalbi and @sbalci.

Are the graphs in the attached screenshot currently possible with your module or with other Jamovi modules out of the box?

Regarding Tidyplot, now integrated into the JASP Plot Builder module, I like the clean, modern color palette; the ability to automatically display p-value comparisons between groups; and the overall layout. I have tested the experimental/alpha modules from @sbalci for Tidyplots, ggstatsplot, and the esquisse package—they look promising.

However, the plots in the screenshots may already be achievable with existing modules. If so, could you point me to the relevant options?

[jonathon]

sure, although it's worth articulating our approach.

that particular plots exist in other stats platforms alone, isn't quite enough for us to think we need to provide them in jamovi. the data we're typically interested in is what proportion of the market are we reaching. that other platforms have taken the time to implement these is certainly one interesting data point, but it's also not everything. the strength of jamovi is as much what it doesn't contain, as what it does.

the first two plots are line plots? which we already provide under the plots tab ... it looks as though there's a few more features available there.

the p-value on box plots seems like another interesting feature ... the question being how much people want to use it. open question. i'm probably in favour of adding them, i just haven't been able to find the data to make the case to the team.

in general, we're more interested in user story, i.e. "these journals insist on plots containing x, y, z", or "its standard practice bioinformatrics to do blah" ... than what plots we might implement.

but these are still interesting examples.

of course, there's also the module system, where others can implement things to their hearts content.

jonathon

[Ravi]

Hi!

Just want to add my two cents. I've looked at the JASP plot builder and I think that using the actual ggplot2 syntax is a lot more user friendly than using this UI layer. The thing is that just copying the functionality of an R package 1-to-1 is not very hard, you just map each R option to a UI option. The hard part is making it user friendly. That means that sometimes you have to think for days/weeks/months/years about what a good implementation looks like. And that often means leaving out functionality so that a first time user can also make sense of it.

Of course, I get that it's nice to be able to make completely custom plots. And our implementation is definitely not finished yet so getting insight into the key functionality that's still missing is great! But we also have to weigh the pros and cons of each addition. In the end, if our implementation is harder to use than just coding it in R, that's pretty pointless to us.

Cheers,
Ravi

[Marielle]

I haven't tried the JASP plot builder yet. But I've used the gamlj way of customizing plots quite often (https://gamlj.github.io/plots_specs.htm ... tomization ) --> run the analysis as usual, copy paste syntax into the Rj module, extract the ggplot object with plot(), then customize with R code.

Common tasks include adjusting the scale, setting axis labels and tick marks, and customizing the legend (content and position).

[vinschger]

It's possible to use the tidyplots library in jamovi using R code.
Thanks to Vincent (@vjalby) for helping me with the installation.

Preparation

1. install R.app

2. In the R console (R.app) install the following libraries
install.packages("tidyplots")
install.packages(„jmvconnect“)
install.packages("tidyverse")
install.packages("ggplot2")
install.packages("ggpubr") —> do not know if this is really needed

3. In Jamovi
- install the Rj module
- select the Rj editor from Rj module menu
- With the wheel icon in Rj window select "System R" and define plot dimensions

Now you can code tidyplots within jamovi.

Example 1:

Code: Select all

library(tidyplots)
study |> 
  tidyplot(x = treatment, y = score, color = treatment) |> 
  add_boxplot() |> 
  add_test_pvalue(ref.group = 1)

Example 2

Code: Select all

library(tidyplots)
study |>
  tidyplot(width = 600, height = 500, unit = 'pt',
  x = score, y = treatment, color = treatment) |>
  add_mean_bar(alpha = 0.3) |>
  add_sem_errorbar() |>
  add_data_points() |>
  theme_tidyplot(font-size = 14)

Example 3

Code: Select all

library(tidyplots)
study |>
  tidyplot(width = 600, height = 500, unit = 'pt',
  x = treatment, y = score, color = treatment) |>
  add_violin() |>
  add_data_points_beeswarm()|>
  theme_tidyplot(fontsize = 14)