Hacker News new | past | comments | ask | show | jobs | submit login

> It is a language, better than MatLab/R/SPSS, which come before it.

I don't think so. I think R is a lot more expressive and not really any harder to read. It might have a steeper learning curve, but it's not so bad that I think that actually matters.

I'm currently learning R. Its a terrible, dreadful programming language. But an excellent DSL for statistical analysis. The main mechanisms for the expressive nature of R is its use of the Environment structure, and terrifying casual use of call stack inspection just to make symbolic APIs. It doesn't even make the latter part of the language without Tidyverse things like rlang. In fact without the Tidyverse efforts the language is even bad as a DSL unless you only deal with a CSV of data with like 100 data points.

I think R is a much worse programming language, like 1 indexing, very unintuitive string manipulation, dataframe being the magic facade that hides complexity, etc.

I would choose Python any day if the other option is R.

1-indexing makes sense within the realm in which R shines. It’s only CS people who seem to completely lose their shit when they encounter it. I’m glad Julia also has 1-indexing.

If you are a computer scientist doing a little data science, then R is awful for the reasons you detailed. If your function entirely consists of data science, then R is a fantastic language for the same reasons you hate it.

R is a fantastic statistical toolkit and a terrible programming language. Building software in it? Massive pain in the ass. Processing tabular datasets on the other hand is an absolute breeze that leaves pandas in the dust. Tidy verse makes it better though.

Yeah, this is generally my take. On the one hand, I cringe when I see all the discussion regarding how much better Python is than R for data science. As a data scientist, who spends 90% of my time cleaning, manipulating, and modeling data, I would take R any day over Python. But at the same time I realize that for many, data science is more 90% software engineering, with 10% being the actual data manipulation and modeling, and for these people, R is a complete non-starter.

Applications are open for YC Winter 2020

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact