R is a language wonderfully suited to scientific programming. The structures and functions available for vector, matrix, and data frame manipulation are powerful and easy to use. Furthermore, the family of libraries making up the tidyverse make working with data frames even easier. The R Markdown file type and R Studio IDE are well-suited for creating reproducible data analysis. R also boasts the best collection of open-source statistical libraries created by researchers around the world. If I wanted to use the most cutting edge statistical techniques, I would choose R.

Now, R is not without its limitations, and I would choose Python over R for some cases. Python has the advantage of being a robust language for web development, with web frameworks such as Flask and Django. Python also has very strong libraries for web parsing, natural language processing, machine learning, and deep learning, and is more generally available on cloud computing services. Python is also more performant than R generally. Python’s pandas library and visualization libraries built upon matplotlib in my mind rival the tidyverse and provide a great alternative.

So generally, I would choose either Python or R for a new data analysis project focused on data explortation, choose R specifically for statistical analysis, and Python for statistical learning and web development.