
Installing the NaN-tb for Octave and Matlab:
--------------------------------------------
a) Extract files and save them in /your/directory/structure/to/NaN/

b) Include the path with one of the following commands: 

        HOME = getenv('HOME');   % if needed, change path to /your/directory/structure/to/

	addpath([HOME,'/NaN/')
	addpath([HOME,'/NaN/inst')
	addpath([HOME,'/NaN/src')

   Make sure the functions in the NaN-toolbox are found before the default functions.
   The NaN-toolbox contains some functions like mean, var, std, cor, cov, and corrcoef 
   which work equivalently than the orginal function - the only exception is that 
   missing values (NaN) are skipped. [You can turn off that behavior by 
   settting flag_implicit_skip_nan(0) ].
   Alternatively, You can avoid this by including the directories at the end of the path
   
	addpath([HOME,'/NaN/','-end')
	addpath([HOME,'/NaN/inst','-end')
	addpath([HOME,'/NaN/src','-end')

c) The use of mex-files is recommended for using the full capabilities of the NaN-toolbox.
   Without the mex-files, SVN- and Liblinear classifiers are not available, and in some cases
   slower m-functions are used. 

   Setup your mex compiler using GCC, the GNU Compiler Collection (or some derivative like MinGW)
      >> mex -setup
   Run MAKE from the directory .../NaN/src/

   Compiling the mex-files is well tested on Linux, but might not work completely on other platforms.
   Specifically, on Windows you need to setup gnumex and gcc (typically mingw or cygwin).
   (covm_mex and sumskipnan_mex are strongly recommended for performance reasons,
   train.mex and svmtrain_mex are needed when using support vector machines) 
   Precompiled binaries are provided for 32bit Windows with Matlab 7.x (tested with 7.1 and 7.6).
   Please note, for Matlab 7.2 or earlier, the pre-compiled mex-files need to be renamed to *.dll; 
   Maybe a *.lnk file with the extension *.dll will also do.  

d) run "naninsttest" from the Octave/Matlab command line prompt 
      >> naninsttest
  In case of success, You should see the following message: 
      >> naninsttest
      NANINSTTEST successful - your NaN-tools are correctly installed

  This checks whether the installation was successful.

-----------------------

  $Id$
  Copyright (c) 2000-2003,2005,2006,2009,2010,2011,2014 by Alois Schloegl <alois.schloegl@gmail.com>
  This is part of the NaN-toolbox
  http://pub.ist.ac.at/~schloegl/matlab/NaN/

