Interactive Projection Pursuit (IPP)

• A Quick Tour

• Some Figures from IPP, Version 2

  1d PP initial projections and a panel of choices	1d PP global options	1d PP investigations with local options	Local options menu
  2d PP initial projections and a panel of choices	2d PP global options	2d PP investigations and other options	More investigations and local options

• A Tiny Tour

• Code and Help page

Interactive Projection Pursuit (IPP) is a program that combines the best features of automatic and manual projection pursuit. The program can be installed as a library for S-Plus, which dynamically loads our C and Fortran algorithms*, for conducting both 1 and 2 dimensional PP; it also provides their (1-d and 2-d) P-values. In addition, they can be used for finding bumps, modes, and clusters in high dimensional data, as well as be linked to locfit. We have tested the program in Splus5 and Splus6 in Linux and Splus3.4 in sunos and SGI. (I do not have Splus6 in Sunos, Splus5 in Sunos is not compatible with our current code.)

• Copyright (c) 2001, Sun and Loader.
  Permission to use, copy, modify, and distribute this software for any purpose without fee is hereby granted, provided that this entire notice is included in all copies of any software which is or includes a copy or modification of this software and in all copies of the supporting documentation for such software. THIS SOFTWARE IS BEING PROVIDED "AS IS", WITHOUT ANY EXPRESS OR IMPLIED WARRANTY. IN PARTICULAR, NEITHER THE AUTHORS NOR LUCENT TECHNOLOGIES NOR CASE WESTERN RESERVE UNIVERSITY MAKE ANY REPRESENTATION OR WARRANTY OF ANY KIND CONCERNING THE MERCHANTABILITY OF THIS SOFTWARE OR ITS FITNESS FOR ANY PARTICULAR PURPOSE.

• Source code: pp.tgz - -- it is a gzipped tar file , save it (pressing the shift button or right-clicking when downloading) in the directory that you are going to install it (either $SHOME/library directory or your own home directory), let's call the directory pp.

• Installation: In the pp directory, type

     tar zxvf pp.tgz
  

  or

     gunzip pp.tgz|tar xvf
  

  depending on what version of "tar" application you have. You'll see install.S6 among many other files. Read install.S6. Basically, to compile in Linux (to use with Splus5 or Splus6), for example, in the pp directory (if it is your own directory, type "Splus CHAPTER" to make it into a S-chapter first, and then) type

      ./install.S6

  If you do not have cc or f77, edit install.S6 to allow for gcc and g77.
  

  To compile in Sunos (to use with Splus3.4), for example, in the pp directory, type

      ./install.S3

  If you do not have f77, change the following lines in install.S3

  cc -c ppind.c
  f77 -c phi.f
  f77 -c pp2d.f
  

  to

  gcc -c ppind.c
  #f77 -c phi.f
  #f77 -c pp2d.f
  

  and save the following compiled binary objective code into your directory: phi.o and pp2d.o and then run the edited installation script install.S3.

• To run ipp in Splus5 or 6, go back to your own home directory, type

  Splus           #or Splus6 or whatever points to your Splus5 or 6
  library(locfit,first=T)  #Optional step - if you have and want to use 
                           #locfit functions inside ipp - it allows
                           #locfit density estimates of projected 
                           #data along interesting projections
  library(pp)     #use library(pp,lib.loc="where.you.put.it") if
                    #pp() is NOT installed globally (ie in the $SHOME)
  pp(x)           #It runs 1 and 2-d ippi on your data matrix "x",
  
  pp(sun1)        #It runs one of our exemplar data called "sun1"
  

• To run ipp in Splus3.4, go back to your own home directory, type

  Splus
  library(pp)     #use library(pp,lib.loc="whereyouputit") alternatively
  motif()         #no need for this in Splus5 or Splus6
  ppa(x)          #It runs 1 and 2-d ippi on your data matrix "x"
  ppa(sun1)       #It runs one of our exemplar data called "sun1"
  

• Exemplar data files: prim7, spiral.std.4, sun1.pp, sun2.ppi, are in the archive of the source code, where spiral.std.4 is from Di Cook and prim7 is from Friedman, available from STATLIB.