Python is not a great language for data science. Part 1: The experience

21 hours ago

Copy Link

Python is considered good but not great for data science, with R often preferred for tasks like data wrangling and visualization.
The author highlights Python's cumbersome nature for quick data analysis tasks compared to R, based on experiences with students and colleagues.
Python excels in deep learning (e.g., PyTorch) but struggles with other data science tasks due to logistical complexities in code.
R's tidyverse is praised for its simplicity and power in data manipulation, contrasting with Python's more verbose pandas syntax.
The article argues for the importance of interactive, low-overhead languages in data science, favoring scripting languages like Python and R.
Performance is secondary to convenience in data science, with the author preferring languages that minimize mental overhead.
The piece critiques Python's need for logistical code (e.g., loops, data type management) versus R's ability to abstract these details.
Despite its flaws, Python remains widely used in data science, partly due to historical accident and its versatility.
The author plans future articles detailing specific issues with Python in data analysis, suggesting deeper architectural problems.

Hasty Briefsbeta