Microsoft maintains an official list of supported R packages available for use in the Power BI Service. There are currently 361 documented packages. While you are not as limited in your package selection when using Power BI Desktop, you have to be more careful if you are working with R in the online Service. If you use a package in Desktop that is not available in Service, the R visual in your report will display an error once published to Service.
Fortunately, the Power BI team lets users know what packages they can use in Service. Unfortunately, from my perspective, the largest disadvantage of the documentation is that there is no description for what each package does.
If you are like me, you may only use a few dozen R packages out of the 10,000+ now available on CRAN. Much of what appears on the list of packages supported in the Power BI Service is a mystery to me. For example, starting at the beginning, there is the abc package. I had no clue that R’s abc package dealt with Approximate Bayesian Computation because I have never encountered that package before. Rather than manually click through the numerous links available to see if each package could be useful to my present work or worth learning about in the future, I built a Power BI report to make things easier.
I used the Power BI documentation web page as a data source and then used the Link field from the supported packages table to query CRAN and obtain the description for each package. Fortunately it is not a report that I need to refresh frequently since it makes 361 separate calls to CRAN.
The description of each package is now more readily available, and I also have the ability to search quickly using the slicer.
If you would like to view the Publish to Web version of the report that includes package descriptions, click here.