think tank forum

technology » r

bluet's avatar
6 years ago
link
bluet
i am learning statistics by following mit's 18.05: https://ocw.mit.edu/courses/mathematics/18-05 … /index.htm

it uses r, which i've never used before. i think some of you have

what are some cool packages, techniques, gotchas for r?
 
6 years ago
link
jason
comes outta nowhere
I use R quite a bit. I work in a lab and often need something to import a bunch of text files, manipulate and analyze the data, and plot it. I'd rather use something open source that is good for plotting, stats, and general data manipulation and R seems like the best option. I like the package ggplot for publication-quality graphs, although the default theme is ugly. Also reshape is good for taking data stored in wide format and converting it to long format. There is a guy named Hadley Wickham who makes a lot of cool packages. ggplot works well with tidy or "long" data frame format which is how data should be organized in R. getting used to the data types and indexing takes a bit of time, but it is great for working with large data sets. I would say to learn the data types. Statistics is a major use for R but if you learn the stats then implementing is trivial, just a command or two. Get used to working with data frames (reshaping, merging, importing, melting/casting) and doing tests on properly organized data is very easy. RStudio is a great program to use for running R with integrated plotting, help, environment, and terminal windows.