Statistics::R (3pm)

Statistics::R is a module to controls the R interpreter (R project for statistical computing: http://www.r-project.org/ <http://www.r-project.org/>). It lets you start R, pass commands to it and retrieve the output. A shared mode allow to have several instances of Statistics::R talk to the same R process.

The current Statistics::R implementation uses pipes (for stdin, stdout and and stderr) to communicate with R. This implementation should be more efficient and reliable than that in previous version, which relied on reading and writing files. As before, this module works on GNU/Linux, \s-1MS\s0 Windows and probably many more systems.

  use Statistics::R;

  # Create a communication bridge with R and start R
  my $R = Statistics::R->new();

  # Run simple R commands
  my $output_file = "file.ps";
  $R->run(qq`postscript("$output_file" , horizontal=FALSE , width=500 , height=500 , pointsize=1)`);
  $R->run(q`plot(c(1, 5, 10), type = "l")`);
  $R->run(q`dev.off()`);

  # Pass and retrieve data (scalars or arrays)
  my $input_value = 1;
  $R->set('x', $input_value);
  $R->run(q`y <- x^2`);
  my $output_value = $R->get('y');
  print "y = $output_value\n";

  $R->stop();

new()

Build a Statistics::R bridge object between Perl and R. Available options are:

r_bin

Specify the full path to R if it is not automatically found. See \s-1INSTALLATION\s0.

shared

Start a shared bridge. When using a shared bridge, several instances of Statistics::R can communicate with the same unique R instance. Example: use Statistics::R;

my $R1 = Statistics::R->new( shared => 1); my $R2 = Statistics::R->new( shared => 1);

$R1->set( 'x', 'pear' ); my $x = $R2->get( 'x' ); print "x = $x\n"; Do not call the stop() method is you still have processes that need to interact with R.

run()

First, start() R if it is not yet running. Then, execute R commands passed as a string and return the output as a string. If your command fails to run in R, an error message will be displayed. Example: my $out = $R->run( q`print( 1 + 2 )` ); If you intend on running many R commands, it may be convenient to pass an array of commands or put multiple commands in an here-doc: # Array of R commands: my $out1 = $R->run( q`a <- 2`, q`b <- 5`, q`c <- a * b`, q`print("ok")` );

# Here-doc with multiple R commands: my $cmds = <<EOF; a <- 2 b <- 5 c <- a * b print('ok') EOF my $out2 = $R->run($cmds); To run commands from a file, see the run_from_file() method. The output you get from run() is the combination of what R would display on the standard output and the standard error, but the order may differ. When loading modules, some may write numerous messages on standard error. You can disable this behavior using the following R command: suppressPackageStartupMessages(library(library_to_load))

run_from_file()

Similar to run() but reads the R commands from the specified file. Internally, this method uses the R source() command to read the file.

set()

Set the value of an R variable (scalar or arrayref). Example: $R->set( 'x', 'pear' ); or $R->set( 'y', [1, 2, 3] );

get()

Get the value of an R variable (scalar or arrayref). Example: my $x = $R->get( 'x' ); # $y is an scalar or my $y = $R->get( 'y' ); # $x is an arrayref

start()

Explicitly start R. Most times, you do not need to do that because the first execution of run() or set() will automatically call start().

stop()

Stop a running instance of R.

restart()

stop() and start() R.

bin()

Get or set the path to the R executable.

is_shared()

Was R started in shared mode?

is_started()

Is R running?

pid()

Return the pid of the running R process

Since Statistics::R relies on R to work, you need to install R first. See this page for downloads, http://www.r-project.org/ <http://www.r-project.org/>. If R is in your \s-1PATH\s0 environment variable, then it should be available from a terminal and be detected automatically by Statistics::R. This means that you don't have to do anything on Linux systems to get Statistics::R working. On Windows systems, in addition to the folders described in \s-1PATH\s0, the usual suspects will be checked for the presence of the R binary, e.g. C:\Program Files\R. If Statistics::R does not find R installation, your last recourse is to specify its full path when calling new():

my $R = Statistics::R->new( r_bin => $fullpath );

You also need to have the following \s-1CPAN\s0 Perl modules installed:

Text::Balanced (>= 1.97)

Regexp::Common

IPC::Run

• Statistics::R::Win32

• Statistics::R::Legacy

• The R-project web site: http://www.r-project.org/ <http://www.r-project.org/>

• Statistics:: modules for Perl: <http://search.cpan.org/search?query=Statistics&mode=module>

Florent Angly <[email protected]> (2011 rewrite)

Graciliano M. P. <[email protected]> (original code)

Brian Cassidy <[email protected]>

This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

All complex software has bugs lurking in it, and this program is no exception. If you find a bug, please report it on the \s-1CPAN\s0 Tracker of Statistics::R: http://rt.cpan.org/Dist/Display.html?Name=Statistics-R <http://rt.cpan.org/Dist/Display.html?Name=Statistics-R>

Bug reports, suggestions and patches are welcome. The Statistics::R code is developed on Github (http://github.com/bricas/statistics-r <http://github.com/bricas/statistics-r>) and is under Git revision control. To get the latest revision, run:

git clone [email protected]:bricas/statistics-r.git