In my last post, I examined the list of R packages supported in Power BI Service. After some additional research, it turns out that I should have titled it “documented” instead of “supported”. Why is that? In addition to the 361 documented packages, there are over 200 undocumented R packages currently available to users in Power BI Service.
I wrote a script for the R visual that uses installed.packages() and puts the package list into a grid using the (documented) gridExtra package. I have not yet tried to obtain the list of the packages as text, but here are two images containing the list of the 600 or so packages currently available (keep in mind that it may not be wise to use an undocumented package in a production environment):
As an example, the beeswarm package does not appear in the documented list, but it is available in Service.
Even though a package is available, it may not be feasible to use it in Power BI. For example, many “interactive” packages with HTML output such as plotly, timevis, DiagrammeR, and more are only available when you create them as a custom visual, not as a self-coded R visual in Power BI Desktop. Others such as RgoogleMaps or RODBC are installed for Power BI Service, but you cannot use them effectively due to your R code running in a sandbox that prohibits outside web calls.
For more detailed considerations and limitations of R in Power BI, see the official documentation.
Finally, for the curious, Power BI Service is currently running version 3.2.2…
On Windows 8 RTM…