RStudio, consider a product rebranding / name change?

Not a question, but a remark about RStudio's current branding and positioning strategy.

In the process of establishing a new data science team and operations for a financial tech startup, I'm running into the usual DevOps/Data Science cultural divide, facing strong prejudices against R as a viable language/ecosystem for production models and client-facing solutions.

Our entire analytical infrastructure relies on (community edition) RStudio Server/Shiny Server/OpenCPU/plumber at the moment (including ETL for business intelligence), and these pieces have performed well so far -- I would only concede that Shiny does not come with standard drilldown/through widgets out of the box, and is not (yet?) a replacement for a BI-centric tool.

Nonetheless after a review of collaborative data science platforms (incl. Domino, Knime, Cloudera, Dataiku, DataRobot, etc.) RStudio Server/Connect did not even make the 1st cut (even though the majority of these platforms actually include RStudio Server). The only reason is that RStudio starts with "R" and that's a plain no-no ("R is for spaghetti code", "R is slow" "You can't do ML with R" "R interface to TensorFlow pales in comparison to Python" "You can't debug R code" "R doesn't play well with AirFlow" etc..).

Not helping either is the fact that RStudio suite does not make it to Gartner's quadrant of Data Science platforms.

I'm very curious if other Community members have had to face similar arguments, and have any advice to share?

Sad to say but a name change and total rebranding would probably help a great deal (-:

Thx!

It seems to me that if the name (rather than the brand or something else) is enough to disqualify RStudio, then something is badly wrong with the evaluation criteria or the evaluators themselves. That's not a particularly productive input, I'm afraid.

If the argument is purely around whether to use R or not in production, there are conference talk videos and blogs (also some threads on here I think) by those who bust these myths. This is one I found from a few years ago, so things will have moved on since: http://blog.sellorm.com/2016/11/26/talk-r-is-production-safe/.

It looks like this is the argument you have to win first.

6 Likes

@martin.R thanks for pointing to Mark Sellors' slides. Seen them before but needed a refresher, excellent.

data.table's ops benchmark is useful as well https://h2oai.github.io/db-benchmark/.

A Who's Who list of mid/large corporations using R in real-time production systems would help as well... not sure I've ever come across one.

1 Like

I would suggest trying the RStudio enterprise tools, in particular RStudio Connect as they really address the majority of the concerns when using R in the enterprise and also make life much easier for developers.

There was a great talk by someone from T-Mobile recently about how they deploy R Models in production and Sean had a fantastic live demonstration on how RStudio Connect can host a shiny app with 1000 concurrent users.

2 Likes

Thanks, T--Mobile Open Source seems to have a series of posts https://opensource.t-mobile.com/blog/posts/r-tensorflow-api/ spot-on.

Indeed, planning to start introducing our team to RStudio Connect on a trial licence.

2 Likes

You may be interested in this list

Also, you'll find testimony at each conference about R. EARL conf is also a conference specific about Enterprise Application of the R Language.

Regarding experience, I can share a bit.

we are using RStudio professional products and very happy about it. I think the main concerns are the R users. Other tools are not enough for our analysts and developers who already have their habits in open solutions with rstudio. The other tools complement or come to fill a gap for other needs. In rstudio suites, The product that have the most value is rstudio connect. It empowers our users in spreading the insights and their data products through the company. I did not found an equivalent for such use.

Other integrated products seems nice but they have their drawbacks too and feels to me more closed than opened. Configuring R inside them and providing a good experience for our R users is very constraint compare to the openness of solution for RStudio.

I won't say it was easy to integrate a R datalab inside our IT system but it really helped toward pushing R day a product into production. The role of analytic admin (which I am lucky to be today) helped a lot into Making R legitimate into the company and bridging the gap between IT folks and Datascients / Analyst.

I am happy to share experience further if you are interested.

Also I think You'll find also some interesting informations about what is possible on
https://solutions.rstudio.com/

1 Like

@cderv, thanks very much. Really appreciate ThinkR-open list and pointers to RStudio Solutions, all very relevant.
Also noted EARL London upcoming Shiny in production workshop.

1 Like