Can you share a repex i.e gist to see how slow it is ?
+1 for highcharter, definitely it deserves a mention, there's a lot one can do with highcharter. In one my blog post I have tried to plot some of my favorite interactive charting libraries in R. https://biranjan.github.io/2017/07/interactive-time-series-plots-in-r/
2 Likes
This looks worth following: @timelyportfolio's https://github.com/timelyportfolio/g2r : experiments with G2, which is "Grammar of Graphics"-based, like ggplot2.
3 Likes
i should warn that these are experimental in nature. I am certainly not in position to build and support a full-featured interactive ggplot2 or ggvis-successor on my own. Happy to start exploring feasibility though.
2 Likes
If anyone wants to join in the fun, I started documenting thoughts and questions in https://github.com/timelyportfolio/g2r/issues. As I mentioned, this is far too big a project to handle on my own, so feel free to contribute.
8 Likes
but dygraph can generate moving averages as well.
It's lot handy for timeseries analysis.
g2 looks really awesome - g2r unfortunately not (yet?)
very alpha experimental g2 htmlwidget built at https://github.com/g2inR/g2r (examples). These are still only sketches to vet various approaches to wrap the API, so the API is very likely to change. I also started an organization https://github.com/g2inR for anyone that wants to join the discussion. So far, one very helpful Matthew de Queljoe has signed up for the possibly very long quest to get this built properly and robustly.
As I progress through the g2 API, I have learned that g2 breadth and scope is far more comprehensive than I initially thought. I should also add that Leland Wilkinson has been advising on the project.
6 Likes
I'm def interested in this @timelyportfolio but am somewhat time-DoS'd until May. I can def contribute at least thoughts/ideas for a while in any org repo convos.
1 Like
Additionally billboarder is similar and (I think) doesn't have the licensing restrictions.
It was used in the Business Science + RStudio Customer Churn Analytics shiny app
https://jjallaire.shinyapps.io/keras-customer-churn/#section-customer-scorecard
4 Likes
The g2r project seems really interesting ! I would gladly help if I can and if I manage to find some time, surely not before february-march unfortunately...
Thanks for the idea and the early work.
1 Like
@juba, @hrbrmstr, and anyone following along Thanks for the interest. With g2r (or whatever we call it) I would like to very carefully consider how we build the API. For instance
-
R6versuslistissue comment? -
+vs%>%vs$(R6) https://github.com/g2inR/g2r/issues/1#issuecomment-359170253
If you have 1-5 minutes, I would l love thoughts.
1 Like
I have a question will get be fast enough to handle at least some middle size data? Second will it have the same Syntex as in ggvis. Because ggvis clearly has the better intuition behind Syntex than ggplot2.
So only 2 questions please do reply
@Anantadinath I am guessing it will be more like ggvis, but I do not intend to mirror ggvis exactly. g2 is canvas-based, so I would expect it to be able to handle middle-size data as quickly as ggvis. When you say "clearly has the better intuition", would it be possible to elaborate?
Thanks so much for you feedback.
In ggplot2
Color can refer to fill in points or the outer edge based on geom that you use. Like in geom points color fills the circles while in geom hist color changes the outer appearance of line.
And you don't have to use rect functions as well.
So on... Ggvis is simple for newcomer ggplot2 is good for people who already know it
I'm curious about the R community's views toward Vega and Vega lite.
For me, Vega and Vega Lite deliver on what I'm looking for from ggvis. They have a coherent vision and architecture, and are very easy to learn if you're familiar with ggplot.
In the Python ecosystem, Vega is available through Altair and pdVega .
Work's also been done in R in the R vegalite package, but it doesn't feel as complete as work in Python. (serious props to AliciaSchep and jsonbecker though.)
Vega is so good that, if there were fully functional R bindings, I'd struggle to see the point in putting more serious effort into ggvis.
3 Likes