I used Matlab for plotting for several years, then switched to R and have been using it for the last few years. I switched because of ggplot2 -- once you get used to syntax (which can take a little while), it becomes far easier to use for exploratory and publication-quality graphing than any other software. For me, the two biggest strengths of ggplot2 are:

(1) "Grammar of Graphics" syntax allows very easy switching between different kinds of figures (lines vs. bars, color vs. black-and-white, etc.). In Matlab I would have to do a lot of code tweaking to switch from a color figure for a talk to a black-and-white version for a paper. The syntax also prevents minor mis-mapping errors (e.g., two conditions have the same color) that can make graphs uninterpretable.

(2) The "stat_summary()" function allows me to use just a single data frame to plot summary statistics (e.g., condition means and SE) or individual participant data or summary statistics for subsets of data (using the subset() function), etc. In most graphing packages I'd have to create a new summary data set or write very complex code each time I wanted to make a new graph, but ggplot2 makes this kind of exploratory graphing very easy.

It did take me some time to learn ggplot2. Hadley Wickham's book (http://ggplot2.org/book/) and the Graphs section of the Cookbook for R website (http://www.cookbook-r.com/Graphs/) are great references. A good place to start is Joe Fruehwald's tutorial workshop: http://www.ling.upenn.edu/~joseff/avml2012/

R is a very good tool for plotting and charting with easy to follow code. Libraries are available for all kind of analyses. Only limitation is memory efficiency, but that is not a factor for smaller datasets.

R is a very good tool for plotting and charting with easy to follow code. Libraries are available for all kind of analyses. Only limitation is memory efficiency, but that is not a factor for smaller datasets.

Octave and R can do the job equally well. At the end it is a matter of personal preference.

R has very powerful visualization capabilities and allow to integrate data analysis with graphical analyses. You can find this guide interesting:

JH Maindonald: Using R for Data Analysis and Graphics: Introduction, Code and Commentary

http://cran.r-project.org/doc/contrib/usingR.pdf

Octave has different capabilities which can be used usually in different contexts than R. The two tools are not strictly competitors but can be used together in a synergy order to obtain better results.

R is quite powerful for multidimensional plotting. Try packages like Lattice by Sarkar.

MATLAB is the most powerfull tool for multidimensional plotting with a big library of ready to use plotting functions. I think that MATLAB plots looks much better than R plots.

R plots look nice bit require pretty much all arguments to be coded. In matlab all objects can be accessed and tweaked within the plot, which is a distinct advantage.

Best wishes,

Carl

everything depends on your knowledge about using some software or programming language. All this are programing languages . IN R you have the ggplot2 library which allows you to do very nice graphics (http://ggplot2.org/)

I used Matlab for plotting for several years, then switched to R and have been using it for the last few years. I switched because of ggplot2 -- once you get used to syntax (which can take a little while), it becomes far easier to use for exploratory and publication-quality graphing than any other software. For me, the two biggest strengths of ggplot2 are:

(1) "Grammar of Graphics" syntax allows very easy switching between different kinds of figures (lines vs. bars, color vs. black-and-white, etc.). In Matlab I would have to do a lot of code tweaking to switch from a color figure for a talk to a black-and-white version for a paper. The syntax also prevents minor mis-mapping errors (e.g., two conditions have the same color) that can make graphs uninterpretable.

(2) The "stat_summary()" function allows me to use just a single data frame to plot summary statistics (e.g., condition means and SE) or individual participant data or summary statistics for subsets of data (using the subset() function), etc. In most graphing packages I'd have to create a new summary data set or write very complex code each time I wanted to make a new graph, but ggplot2 makes this kind of exploratory graphing very easy.

It did take me some time to learn ggplot2. Hadley Wickham's book (http://ggplot2.org/book/) and the Graphs section of the Cookbook for R website (http://www.cookbook-r.com/Graphs/) are great references. A good place to start is Joe Fruehwald's tutorial workshop: http://www.ling.upenn.edu/~joseff/avml2012/

I agree with Richard, that both software will get the job done - so aesthetically it depends on what you prefer - personally I choose R. However keep in mind that R requires extra coding and in my opinion is worse documented.

Getting back to the initial question: R/Matlab/Octave/others do the same job similar.

All have advantages and drawbacks. It often depends on the specific question. However, I prefer R in most cases. It is free, free packages are available and it is great for students, as they can simply download and customise their own version.

Maybe you find this web Ressource helpful. I wrote it a couple of years ago:

http://www.code10.info/index.php?option=com_content&view=article&id=79:articler-basics-and-statistics&catid=59:code-a-programming&Itemid=82

Jan

Thanks a lot all for providing insights and your experienced comments minimized the dilemma.

I would go with R. It produced the best plots and is very flexible and remember to use RStudio... Its also a good oppurtunity to gain familarity with R as the software increases in popularity.

MATLAB is very good also especially if you have a lot of expereince with it.

I agree. R is great and it is a great tool for anyone doing data analysis.

I was recently writing some book chapter, pre-existing parts were in MSWord, so that was the requirement. Exporting graphs from Matlab, I was not able to achieve satisfactory publication quality vector graphs (sure, could be lack of skills on my side). Excel did not make me happy either. I ended up programming graphs in R and exporting them in PDF format.

..I would characterise myself as more skilled in Matlab than in R, which I have been using now and then for some 10 months.

Matlab is the most powerful as already mentioned, but if you're not perfect in this language and/or have to do the analysis fast and simple it's not the perfect choice for you.

For pure statistical analysis you should use SPSS or if you don't know how to use it / have no one around who's a experienced SPSS user... Then you should use R.

At https://www.codeschool.com/courses/try-r you can find some basics about R.

BTW, in this context, you may find interesting this book about exploratory data analysis in Matlab: http://www.crcpress.com/product/isbn/9781439812204

I recommend R, a free language that makes it easy to get reasonable graphs and which offers the wide world of statistical analysis as well.

As has been pointed out, it's easy to incorporate R graphics into text preparation software, a task that can be tricky with Matlab. Another downside of Matlab is its cost. As for Octave, bear in mind that it is not fully compatible with Matlab.

ggplot2 is really a very good package for graphics with high quality. One needs to familiarize himself of its syntax. I used it lot for my GLM course assignment.

They'll both work so it's really a thing of syntax/community preference.

I tend to find there is more online help for R, although the R syntax can get really weird (especially if you also use other languages like vb, fortran, c#, etc). but overall I like R.

I'd disagree with Daniel's suggestion that you should use SPSS for pure statistical analysis (R is free and much more flexible, besibes for basic stats you can type in R code faster than clicking away in SPSS).

If you want to go with R, I highly recommend RStudio - it's the IDE that makes the difference :) http://www.rstudio.com/

I am not sure that plain R is much better for graphics than Matlab. The last time I tried to use Octave (on Mac OS), plotting required further installs that took me a lot of time, but it may have improved since then (about 1yr ago). I would certainly recommend R, with H. Wickam's ggplot2 package. Note however that learning to use it efficiently takes

time.

I think R is really better than others ! Free and customizable with a real help community, full implemented packages, all platforms and, with experiment, outputs look very fine, particulary if you work with latex (cf tikzDevice package) ! I discover all days new functionnalities in R ;-)

you can make very good prototypes and learn about ggplot by using the web app at http://rweb.stat.ucla.edu/ggplot2/. This is a great way to start.

Any software could be appropriate, it depends on your knowledge on it. However I think that Octave does not exactly match Matlab capabilities and could be a litlle bit deprecated.

Hope this helps

I just wanted to follow up on @Ashok Sharma's answer. I agree ggplot is very powerful. For those who have MATLAB experience and might find the task of learning R daunting there is a package that emulates many of ggplot's functions in MATLAB called gramm (as Pierre eluded to). I've used it with great success when I have large datasets that would bog down in R.

You can view it here

https://www.mathworks.com/matlabcentral/fileexchange/54465-gramm--complete-data-visualization-toolbox--ggplot2-r-like-