none
R langage RRS feed

All replies

  • R is rather scripting {interpreted} language than programming{compiled} (although in both you are implementing workflow/logic).

    Also on a market Python does dominate significantly over R while offering similar functionality
    (package/lib based extensions, despite initially targeting/approaching from different ends - Python from ETL side, R from Statistics/Numeric-Math scope, now both have converged to provide same functionality) - it is evidence-confirmed by both: market size and dynamics (incl. native scalable platform integration/support)

    There's still an issue with R on RAM limits (in a way addressed, differently depending on package)

    I wouldn't bet much on R future (for inst. vs Python {e.g. Scala/Databricks/Kaggle, etc}) because of relatively low market share (MS has tried with some acquisitions and marketing with strong product integration {Azure Cloud range: Machine-Learning / Data-Science and SQL Server product class, Power BI, etc.} but looks like gradually giving up (adding support for other) because reality check suggests these efforts might be an expensive waste of resources into the void/dead-end over longer period of time), meaning your knowledge/experience on R very likely to become obsolete / not demanded much on a labor-market.

    Friday, July 3, 2020 11:01 AM
  • I come at this from a different angle, being a health sciences professional (doctor) who is not planning on earning money from programming (well, a few dollars won't hurt...)

    I also came at this from a situation where I used Python for everything, and was somewhat disappointed that the data science courses I was interested in used R, not Python.

    Particularly the *applied* data science and post-graduate university courses for non-programmers who want to analyze their own data.

    I can now see the considerable benefits that R offers those with an interest in cleaning and analyzing data compared to Python. At least for a part-time programmer. And perhaps for those from a more 'science' background who need to do proof-of-concept exercises before passing on the programming task to a 'production'/engineer team (who may well prefer Python...).

    So, perhaps, I agree that big-time production code may well end up being done in a more production-friendly language. Rapid prototyping, data cleaning, exploratory data analysis and 'one-off' analysis (reproducible for scientific validity, unlike shuffling numbers around an Excel spreadsheet, but not intended to be repeated ad infinitum on fresh sets of data in perpetuity) is where I find R useful, and it sounds like R's place in the programming world.

    Saturday, July 4, 2020 3:13 AM
  • It depends on what you want to use it for.

    As Yuri says, R is used less in industry compared to Python, however in the various sciences and academia, especially biological (my background), R is by far the dominant language used for any type of analysis with no signs of shifts towards python. Many datasets e.g. eBird have filtering and analysis tools only in R, and analysis is much easier in R being able to run line-by-line, as well as way easier to find bugs.

    If you intend on going the academic route, R is absolutely worth learning. If not, less so, but its still a useful skill.

    Tuesday, August 4, 2020 11:18 PM