Vignette interactivity

[casperhart]

• Remove antialiasing logic due to weird edge artifacts. Also modify depthTest value to ensure points are drawn in the correct order.
• Fix clearing of selection when crosstalk isn't in use.
• Vignette on interacting with the display_scatter widget.
• Add pdfsense dataset.

@earowang I've published the vignette in this PR here: https://casperhart.github.io/detourr/articles/display_scatter_interactivity/display_scatter_interactivity.html

I also haven't documented the pdfsense dataset yet or added any attribution. I'll look in to what's required for this tomorrow.

[earowang]

All icons are invisible but clickable when the widget displays on a mobile device.

[casperhart]

I'll create a separate ticket for mobile interactivity. Most of it should be easy enough, but the selection controls might require some work

[earowang]

@dicook thinks the variables shown for pdfsense are not quite right - there shouldn’t be the big gap - Variables 7 through to 62, but use for 6 PCs of a PCA

@casperhart Can you double check on this?

[dicook]

Something like this code to pre-process: data(pdfsense) # play around with the gamma parameter pcs <- prcomp(pdfsense[, 7:ncol(pdfsense)]) pdfsense <- bind_cols( pdfsense, as.data.frame(pcs$x) ) # This is the class variable to use for colour pdfsense$Type <- factor(pdfsense$Type) # Standardise all to have same scale pdfsense_rs <- select(pdfsense, PC1:PC6) %>% mutate_all(.funs = function(x) (x - mean(x)) / sd(x))

On 15 Feb 2022, at 11:14 am, Earo Wang ***@***.***> wrote: @dicook thinks the variables shown for pdfsense are not quite right - there shouldn’t be the big gap - Variables 7 through to 62, but use for 6 PCs of a PCA @casperhart Can you double check on this? — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you were mentioned.

cheers, Di

…

--- Dianne Cook ***@***.***

[casperhart]

The issue with the gap in the plot was that the variables for the label aesthetic were also being used in the tour itself, e.g. mu, InFit. A rookie mistake I know. I should be able to fix up all the screenshots tomorrow to match the updated display.

I've also found the issue with the icons not showing on mobile. I've gotten them to show, but the controls will require some work for touch screen. That will be covered in #89.

When documenting the pdfsense dataset, I'm not sure whether to include the data in the package directly or have {liminal} in the package Suggests. If using the data directly in {detourr}, is it okay to reuse the same documentation as {liminal}?

[earowang]

Should suggest {liminal}, and call data(pdfsense, package = "liminal") to use it.

[casperhart]

I've addressed all the review comments so far, let me know if any other changes are needed