Recs.
Updated
SpecsUpdate
Pros
Pro Lots of packages available
There are lots of different packages available that can be easily searched from the CRAN repo site and downloaded, or installed via the R command line interface. These packages are easy to include in a project or source file, and pertain to a wide variety of topics, from classification, social media analysis, and text processing to interactive 3d plotting and networks (including neural nets).
Pro Has a wide range of options when it comes to IDEs/GUIs
Among the IDEs available there are several commercial applications as well as free and/or open-source ones, such as R Studio, which features syntax highlighting, project management capabilities, integrated terminal access, decent code completion and on-the-spot parameter hinting, graphical interfaces for package installation and such, and commendable extensibility/developer support.
Pro A variety of implementations for all possible algorithms and statistical methods (nowhere else found)
Pro Microsoft is now involved
Which means more high quality tooling (support for R in Visual Studio) & libs & integrations (e.g. R in SQLServer) & cloud solutions for scalability beyond single core & single host (R server , Spark via sparkR/sparklyr lib,... all on Azure supported by MS) .
It also means a lot more 'mainstream' developers are going to be exposed to R (e.g. via SQLServer) and hence R is going to become a lot more popular outside of its traditional academic statistics niche
Cons
Con The library ecosystem (mostly CRAN but also github, BioConductor & others) is rather fragmented
There are over 10000 R packages and rising and there are many packages with overlapping functionality. It's not always clear which package is "best" for your needs. The situation is getting better thanks to Hadley Wickham who has rewritten a lot of core functionality from R base & popular but 'old school' packages in a few big alternative packages in a new more modern style and also more coherent (the so called 'Hadleyverse' or 'Tidyverse'). Together these form a framework analogous to what you have in Pydata for data science in Python or Spring for web applications in Java.