R or Python, that is the question!

I want to answer the following questions that I get a lot of, especially, in my data science classes and even from data analysts and developers that have been in the field for some years.

• R or Python?
• Functional programming language or interpreted language??
• Statisticians or software engineers?

So, I want to tackle these questions really quickly in one sentence:

R and Python are both just tools! use whatever language you like better!

We know that R is more data science software than Python which is a general-purpose object-oriented programming language. Python is more aimed at software engineers and happens to support data science as one of its applications. However, as a data scientist you must learn both, because the choice between Python and R will depend on the specific project and the skills and preferences of the team and you must be ready. Many data scientists use both languages and choose the one that best suits the task at hand. In fact, R and Python are complementary and there is no significant difference between them.?

Why?

1. In code efficiency and speed, there is no meaningful difference as most modern libraries in R and Python are now written in C/C++!
2. In security, some people are saying Python is safer especially as a back-end server for calculations over the web. This is true but not a serious security vulnerability! Because you always can use virtual containers on the cloud! Additionally, Python also suffers from security problems such as SQL injections in NLP projects. It is very difficult to approve or reject one of them for security issues! Both are suffering from security problems and all problems have their special solutions.
3. In visualization, R is great! Few lines of code and much easier to customize!
4. In packages and standard libraries, R has over 13,000 standard packages for data analysis, manipulation, and visualization, and Python has 200 standard libraries. Additionally, remember Python upgrading problems from 3.7 to 3.11 and backward incompatibility of Python 3.0 with version 2.0!
5. In coding, R needs much fewer lines of code for the same result in comparison with Python.
6. In statistical concepts, R handles categorical data much better and easier than Python.
7. Finally, Python lovers might think R is not very efficient in some advanced machine learning tasks such as web scraping, MLops, or deep learning! It is absolutely wrong! There is literally no task that can be done on Python that you can't do on R. Only look at the following strong R packages:

• For Data engineering (Data.table,?tidyverse),
• For Webscraping (RSelenium, rvest),
• For Web/mobile application development (Shiny which offers much more flexibility and functionality than full dashboarding solutions built with Python such as streamlit, dash or even flask),
• For API development (Plumber),
• For Go to production (Rocker, Plumber),
• For MLops (Vetiver),
• For Advanced machine learning or deep learning (Tidymodels, torchR, keras, tensorflow),
• For Computer vision (Rvision, torchvision),
• For NLP natural language processing?(spacyr, TM, KoRpus, openNLP, wordcloud, tidytext, and text2vec),
• and for streaming data from servers to PCs (parkR, sparklyr) similar to PySpark.

I came from SPSS/SAS/Splus/Minitab/Stata/Eviews/Matlab in undergrad(1996) and worked with R starting in 2002, Python in 2010, and PySpark in 2016. I am certified in Network+, Cisco Routing Switching (CCNA), Windows Server (Microsoft Active Directory), and LPIC for Linux, and I started coding in 1996. All of them with no documentation, no StackOverflow, and no GitHub repos. Only one reference book for each and some classes!

Nowadays, we have many resources available online! If you're just beginning or are a few years into your data journey, keep going with continuous learning and growth! Start with R if you like to be a data analyst or Python if you are a programmer, simply by enjoying whatever is shared with their powerful communities! You might need to use only one of these tools according to the workflow or combine them according to your final product. Remember that they are only tools and there are many others in the market! Focus more on data science concepts rather than the software and tools, and never participate in the Python vs. R debates!

#R #Python #datascience

Q/As:

Our discussion group: https://www.dhirubhai.net/groups/12420006/

Q1: According to DZone, the popularity of Python has grown much more than R's. Could this indicate that in the future Python will have more libraries than R? Is this rise in Python's popularity driven by business?

A1: In my opinion, Python is a great option for data analysis and learning the computer programming logic. If you like programming, go for Python and remember R and Python are the only vehicles to get to destination. Regarding popularity, I guess all these charts and online trends are not accurate. Why? (1) For R before 2018 as the most popular one and after that Python took over, the population, purpose, or business sector of these comparisons are not clear! (2) More online activity or search keywords are not popularity! (3) High job demand is not also all about software! Companies might prefer employees with data scientist skills that could also work as a programmer, a developer, and an engineer! They want all-in-one! Their strategy is Python might be better because it is a multipurpose language! Of course, Python might cover all of these but almost impossible for a human being! Data science is teamwork, and software never matters for a professional team! Historically, we could see different programming languages and professionals with diverse skills in one project! Knowledge of data science concepts is more important than its tools, and in my opinion, AI will manage the software part soon. So, use whatever language you like better!

Q2: I have heard that R libraries are often times peer-reviewed or not reviewed at all. Do you believe it to be a limitation of R and could that constitute a hindrance in terms of fact checking results in data analysis?

A2: Unfortunately, there is fake news about R and Python! Let me give you some resources. We have the following organization for Peer review of R packages. rOpenSci for R: https://ropensci.org/packages/all/ . We also have PyOpenSci for Python (PyOpenSci was modeled after rOpenSci). The first package submitted to pyOpenSci was in May 2019. pyOpenSci for Python: https://www.pyopensci.org/python-packages.html . You can check the peer-reviewed packages in both. For example, in 2019, we had 12,500 R packages. More or less 6000 beta version packages. It means we still had 6500 final R packages! Also, check the number of R packages available on CRAN and Python online package repository PyPI. These numbers were discussed only because of the question but remember nobody cares these numbers in R or Python communities! In Python, you can do most of your job only by NumPy and pandas! My point is the knowledge of data science concepts is more important than its tools. So, use whatever language you like better and remember, R and Python are complementary, and there is no significant difference between them! Finally, two good references if you are interested in this topic: Link1 , Link2

Summary of comments (Last update: 16/02/2023)

After exploring the use of both R and Python in financial data research, it has become clear that each language has its advantages and disadvantages. The optimal choice depends on the specific task and the user's preferences. R is highly suited to statistical modelling, analysis, and data visualization, with a robust ecosystem of libraries specifically designed for financial analysis. Python is a general-purpose programming language that offers various libraries for numerical calculus and data manipulation.

Some comments show that Python may seem cleaner and more intuitive for newcomers to programming languages, but R is highly efficient for plotting data with simple and concise code. Additionally, novice programmers often prefer Python because its code and documentation are more user-friendly. However, R can perform complex operations in a single line of code. In contrast, Python requires the user to spell out each step, which may be tedious for experienced programmers. R and Python have advantages and disadvantages, but individuals aspiring to become data scientists should learn both languages.

Using packages in R or Python enables the central aspects of algorithms and details to be incorporated into the code, eliminating the need for step-by-step programming. However, the comments emphasize that individuals must possess the required theoretical knowledge to use these packages effectively.

During the post-graduation in Data Science for Finance, students have gained knowledge of R and Python. Some find it easy to explore new solutions for these open-source programming languages online. With R, packages are available online to help with the analysis, while with Python, one needs to create some structures from scratch.

Some comments show students' interest in coding in Python for data manipulation and visualizing data using R. They also agree that many programmers prefer Python, but data scientists are not necessarily programmers. Experts often contribute to both communities to create better tools.?

Some comments show an interest in coding in SAS, Python, Spark, R, and SQL, which students found necessary for their different projects. They also mentioned MATLAB as another valuable tool in this field.

Some comments mentioned that Python has a higher preference in the worldwide market, but no concrete evidence or scientific study exists.