I generally reach for Python when building data science pipelines, however I discovered R before I decided to invest in learning Python. R has saved me lots of time when it came to quickly and easily preparing nice-looking plots for research. It begs the question of where is R better than Python for certain purposes? We will discuss the benefits and downsides of Python and R so that we can reach for the appropriate tool when needed, instead of treating all problems like nails when a screwdriver is required.
Ultimately, anyone could build data science workflows exclusively in R or in Python with enough time/expertise, but there are certain use cases where one language may be advantageous over the other. Before choosing a tool, individuals should consider:
- Tools commonly used in their field – this may dictate availability of repos or external forums that can be consulted for help toward getting the task done efficiently.
- Tools used by colleagues/team – using the same tools means that code can be shared and simpler tool stack overall.
- Cost of learning a new language – it can take years to effectively learn a new language that solves the problem.
| R | Python |
| Great for stats, data analysis/exploration and visualization. Interactive & aesthetically pleasing graphics/dashboards. | Seaborn and plotly are great data visualization libraries with similar (but fewer) plotting capabilities. Visuals are not as informative. (This is rapidly changing!) |
| Excellent Reporting tools e.g. RMarkdown (multimedia, journal quality report) and Shiny (fast prototyping of interactive web apps). | Has Jupyter Notebooks, but capabilities offered are limited compared to R. No Shiny equivalent. |
| Rooted in Stats, developed for researchers and scientists to design, perform and communicate data analysis results. | General purpose with roots in computer science and math. Useful for tasks beyond data analysis. |
| Catching up to Python in this space | Excellent data science libraries for NLP and deep learning e.g. TensorFlow, scikit-learn, and web frameworks libraries for scripting websites at scale. |
| Better for Web scraping and crawling, and database connections. | |
| Faster computations and import of large input files. Good for mathematical computation and understanding algorithms | |
| Production – ready language that can integrate all parts of the workflow. High ease of deployment and reproducibility. | |
| Easy to learn for folks with little to no coding experience, but more difficult to develop expertise in due to advanced functionality. | Folks with a computer engineering background will find Python easier to learn. It has a linear learning curve. |
| Many ways of writing the same functionality. | More consistency in the way code is written (i.e. pythonic way of writing code). |
Python leads R in overall popularity and is projected to completely overtake R in terms of usage in the future, making it the safer tool to use going forward.
R is great for early exposure to coding and for learning statistics especially while taking a course due to its extensive statistical libraries. However, as a budding data scientist builds their coding expertise and moves toward creating production-ready machine learning models, Python becomes the better choice.