MultiClass Visualizable Classification using Combination of Projections

An ensemble classifier for multiclass classification. This is a novel classifier that natively works as an ensemble. It projects data on a large number of matrices, and uses very simple classifiers on each of these projections. The results are then combined, ideally via Dempster-Shafer Calculus.


News

Version 1.3.2 Wrote a longer description line for the package, to more fully describe what the package does. Hat Tip: Uwe Liggs

Version 1.3.1 parallel processing now removed from ensemble examples to pass CRAN check Namespace Imports from stats package implemented
Hat Tip for all of these: Kurt Hornik Also cleaned up main package documentation file

Version 1.3 Belief Builder functions' options are now implemented as a parameter list Documentation updated to reflect this change A proper Ensemble function has finally been written Parallel processing fixed in the ensemble function Ensemble function now has a function to return all the classification matrices in order Package now depends on 'parallel' package

Version 1.2 Added Full namespace export directive Removed dependency on cepp -Added files bases.R (Contains 3 functions from the tourr package by Hadley Wickham) -Added original documentation file for basis_random Reorganized files -combination functions moved to file combine.R -documentation moved to combine.Rd Added a verbose option to bpa and bpamat print objects Rewrote combine.ds function

Version 1.1 Is a Major Upgrade from before, primarily in terms of architecture get.NN Can now return the largest distance it finds (This feature used in the jitter belief function) Conversion from p to k now automatically handled inside get.NN function Functions least.p and which.least.p still retained in file knn_header.R Documentation Updated to include examples of new feature bpa Now declared as a class, with print method & Documentation Updated bpa.mat Now declared as a class, with print method & Documentation Updated. Example Added. Removed file DS_Classify.R, replaced with Belief.R New Belief Functions documented Documentation file for proj_classify removed File Ensemble.R removed (temporarily) Documentation file ensemble.Rd removed (temporarily)

Version 1.0 Combination function was coded wrong, retain, but code new DS function Done & Verified for bpa objects Done & Verified for bpamat objects Rewrite assigned ratios function so that it always returns ratios to ignorance

Version 0.9 Fix bug in get.NN function - distances were being returned wrong
Add more belief functions (New File: Belief.R)

Version 0.8 Revert proj_classify to that in package version 0.5 Remove 'info' argument from proj_classify Update proj_classify documentation Fix bug in ensemble function where nmats was not being updated Update ensemble function to work with the new proj_classify function Document the ensemble function

Version 0.7 Including an ensemble function Fixed a bug in the bpamat combine function

Version 0.6 proj_classify now returns bpa.mat type objects Documentation of proj_classify and bpa.mat updated to reflect this change

Version 0.5 proj_classify now accounts for ignorance in classification proj_classify now has a field for info Fixed a bug in proj_classify where the row names were being assigned wrong

Version 0.4 A new R file is added: DS_Support.R Allows storing multiple bpa objects as matrices Meant to better support DS operations in classification bpa objects now have two additional fields - name and info bpa objects now return full classification info Relevant Documentation is added / updated

Version 0.3 Package compiles without error or warnings proj_classify now has a working example Breast Cancer data is now documented

Version 0.2 UCI Breast Cancer dataset is now part of the package bpa objects now return classification information bpa lists can be combined at once proj_classify is now properly documented

Version 0.1 First Package Version

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("MuViCP")

1.3.2 by Mohit Dayal, 3 years ago


Browse source code at https://github.com/cran/MuViCP


Authors: Mohit Dayal


Documentation:   PDF Manual  


GPL-3 license


Imports sm, MASS, gtools, parallel


See at CRAN