Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Data Analysis and Visualization Using R (2014)

varianceexplained.org

Post reply on HN
Print view View original

11–20 of 26 posts

Re: Data Analysis and Visualization Using R (2014)

#11

Post by stewbrew »

sgt101 wrote: post #9

I love R, but I have two problems with it that I would like suggestions to deal with. 1. Debugging seems way more primitive than in other languages; I get cryptic messages and really struggle to pinpoint what is happening. Debugging in (free) shiny is even harder, the page says connection closed and I have to guess what has happened. 2) Code structure. R is simply fantastic in REPL and/or RStudio mode for digging aro…

Ad 1. See ?browser ?traceback ?debugger. You also have breakpoints.

Ad 2. There are reference classes which also provide limited type checking for fields. You can also encapsule your code in environments which is more R-style but doesn't work well with roxygen.

Re: Data Analysis and Visualization Using R (2014)

#12

Post by jcheng »

sgt101 wrote: post #9

I love R, but I have two problems with it that I would like suggestions to deal with. 1. Debugging seems way more primitive than in other languages; I get cryptic messages and really struggle to pinpoint what is happening. Debugging in (free) shiny is even harder, the page says connection closed and I have to guess what has happened. 2) Code structure. R is simply fantastic in REPL and/or RStudio mode for digging aro…

FYI, debugging in Shiny has gotten much better in 0.13.1 and later (you now get stack traces at the console, or in your log file if running on your own Shiny Server, or in your admin console if running on ShinyApps.io).

Re: Data Analysis and Visualization Using R (2014)

#14

Post by sgt101 »

jcheng wrote: post #12
sgt101 wrote: post #9

I love R, but I have two problems with it that I would like suggestions to deal with. 1. Debugging seems way more primitive than in other languages; I get cryptic messages and really struggle to pinpoint what is happening. Debugging in (free) shiny is even harder, the page says connection closed and I have to guess what has happened. 2) Code structure. R is simply fantastic in REPL and/or RStudio mode for digging aro…

FYI, debugging in Shiny has gotten much better in 0.13.1 and later (you now get stack traces at the console, or in your log file if running on your own Shiny Server, or in your admin console if running on ShinyApps.io).

How? I have looked, but shiny says "debugging is private, talk you your admin".... (I'm running ubuntu on aws)

Re: Data Analysis and Visualization Using R (2014)

#15

Post by sgt101 »

blahi wrote: post #10
sgt101 wrote: post #9

I love R, but I have two problems with it that I would like suggestions to deal with. 1. Debugging seems way more primitive than in other languages; I get cryptic messages and really struggle to pinpoint what is happening. Debugging in (free) shiny is even harder, the page says connection closed and I have to guess what has happened. 2) Code structure. R is simply fantastic in REPL and/or RStudio mode for digging aro…

Sounds like you need some Visual Studio in your life. And you probably need to be more assertive. https://cran.r-project.org/web/packages/assertive/index.html https://www.youtube.com/watch?v=JWjiMvlfCwk (and you do use testhat for unit testing, right?) Additionally you want to write more modular code. There is lots of infrastructure around that in R, but people just don't use it often enough because a lot of them are…

I use r-studio but not sure what visual studio will bring - will investigate.

Re: Data Analysis and Visualization Using R (2014)

#16

Post by blahi »

sgt101 wrote: post #15
blahi wrote: post #10

Earlier quoted context omitted.

Sounds like you need some Visual Studio in your life. And you probably need to be more assertive. https://cran.r-project.org/web/packages/assertive/index.html https://www.youtube.com/watch?v=JWjiMvlfCwk (and you do use testhat for unit testing, right?) Additionally you want to write more modular code. There is lots of infrastructure around that in R, but people just don't use it often enough because a lot of them are…

I use r-studio but not sure what visual studio will bring - will investigate.

Project management and develops tools.

Re: Data Analysis and Visualization Using R (2014)

#18

Post by denzil_correa »

blahi wrote: post #17
denzil_correa wrote: post #13

How good is the support for R when data is large and does not fit in memory?

plenty of options, depending on what you need.

I guess one has options in every language including Python. Does something make R stand out over Python for Big Data?

Re: Data Analysis and Visualization Using R (2014)

#19

Post by vegabook »

I use R multiple hours every day, I love it, I love the ecosystem, but I can't help thinking that it's showing its age. It does a great job on static analysis and visualization of smaller (< 1 gigabyte) data sets, but is seriously challenged by anything significantly larger, and is unfit for purpose if the data is changing rapidly (eg streaming). I unfortunately am slowly coming to the conclusion that Spark and Flink style tools are where data science will be at in a few years time, and while I know you can use R as a layer on top of these, I think other aspects of R also hold you back, paradoxically, things like the excellent base and ggplot graphics, which are rightly lauded as excellent, but are very low-dimensional in a world where tensors increasingly rule. I think R will remain hugely relevant for a long time, and is, what I tell people, like Excel^2, but it's getting to point where the world is moving on and it will struggle if it's not rewritten from the ground up with a much faster, multicore, threaded, distributed implementation.

Re: Data Analysis and Visualization Using R (2014)

#20

Post by blahi »

denzil_correa wrote: post #18
blahi wrote: post #17

Earlier quoted context omitted.

plenty of options, depending on what you need.

I guess one has options in every language including Python. Does something make R stand out over Python for Big Data?

Everything that makes it stand over Python for small data.
Post reply on HN