R

[Predrag Punosevac]

Kirthevasan Kandasamy <kandasamy at cmu.edu> wrote:

> alternatively,
> you could switch to matlab :P
> 

That is a bad advise! I am serious. 

As you know both MATLAB and R are domain specific languages. The only
data structure recognized by MATLAB is matrix.  Two dimensional matrix
of a finite subset of rational numbers. How are you going to put a
string into the matrix?  Ugly yeah!

On the another hand basic data types in R are:

- Numeric
- Integer
- Complex
- Logical
- Character


which means that with exception of pointers R recognizes all classic
primitive data types plus a buil-in data type "Complex" (Fortran lovers
will remember that one). R has more packages for data mining and
machine learning than any other language and this comes from a mouth of
a Pythonista. 


Now why some nut cases like Peter and me like Python better than R. Well
unlike R, Python is general purpose interpreted language which has two
great libraries for dealing with data mining and machine learning Pandas
and Scikit-learn. They are not quite as good as R (they are getting
there) but we are completely unconstrained by the domain of application
so important to R.


We could also talk about other alternatives like Perl, Haskell, Julia,
Lua alike.

Cheers,
Predrag


> On Tue, Sep 29, 2015 at 8:35 PM, Predrag Punosevac <predragp at cs.cmu.edu>
> wrote:
> 
> > Maria De Arteaga Gonzalez <mdeartea at andrew.cmu.edu> wrote:
> >
> > > Hi Predrag,
> > >
> > > My desktop has R 3.2 installed, however, many of the packages that I use
> > for my research only work for R 3.1, could I downgrade? Thanks,
> > >
> >
> > Actually after trying to downgrade your version of R for about 45
> > minutes I realized that is a wrong way to deal with the problem.
> > Downgrading packages is very difficult as all dependencies have to be
> > resolved manually. In system administration we call that Red Hat Hell.
> > The real way to deal with your problem is to upgrade manually installed
> > package. Kyle brought to my attention that automatics updates of R
> > (which I turned on)  are braking manually installed R packages. There is
> > a way to deal with it. I just need to read about 20 pages of the
> > documentation before I can do it.
> >
> > Between plotting in R shell is broken since last week due to shady QA by
> > EPEL people. I am hoping that the upstream will fix the problem soon and
> > that will not have to "downgrade" things.
> >
> > Finally Dr. Dubrawski have overheard me using word obsolete in the same
> > sentence with R. Our R is the latest and the greatest (or maybe not so
> > great). Our installation of RStudio (IDE) is obsolete as upstream cares
> > only for Windows. The upshot of having the old version of RStudio is
> > that R plots still work from it.
> >
> > Best,
> > Predrag
> >
> >
> >
> > > Maria
> > >
> > > Mar?a De Arteaga
> > > PhD Student in Machine Learning and Public Policy
> > > Carnegie Mellon University
> >
> >