Approach from the general to the specific.
IoT is telemetry. Signals travel from one device to another, either one way (turn on/turn off) or two way(turn on/return image). The program handling the communication runs according to a protocol that can be programmed against. Decide whether to give over to proprietary protocols or open.
Protocols can be implemented variously. A lowest common denominator approach has benefits if the universe of devices is known. csv works everywhere. If the universe is unknown, xml acts as a self-documenting data format.
The data store in which to capture depends on volume and frequency. Appending to csv is a standard i/o operation, but is tricky to handle when there are race conditions (two devices trying to write at the same time). That is and large volumes are when SQL is indicated. It is also useful when data validation is required.
Give data in storage, the volume influences the choice of analytic software. R is limited by memory; it has to have the data to be processed in memory at the same time. This is only relevant with very large datasets.
The scope of statistical analysis determines whether R should be considered over Python. Basic descriptive statistics can be done in either. R has an advantage in advanced statistics.
R has more expressive graphing capabilities than Python. If the finish of presentation is important, that argues in favor of R.
Learning two languages at the same time can be difficult if the languages dififer greatly in philosophy. Just learning Russian and Ukrainian (C and Ruby) is easier than learning Basque and Japanese(R and Python). Python is a procedural/imperative language: do this, do that, and then do the other. R is functional, f(x) = y. Having Python as a first language can make it difficult to pick up R; less so, v.v.
The subsequent development of a project created in a scripting language, such as R or Python to scale to large demands, may require compiled language implementation. It is generally easier to migrate Python to C++, say, than R to Go or Haskell.
Specific questions.
- Undergraduate level business statistics is an R built in. Beyond that depends on the specific questions to be put to particular data.
- RStudio is a GUI to R; database tables as data frames are native to R and extensions, such as tibble enhance them.
- R is indifferent to what happened to the data before arriving in a readable format, such as csv or SQL.
- There are some R projects specifically aimed at IoT, such as the iotables package.
- Keep in mind that RStudio can be used as a Python IDE