Matlab
codes for principal component accumulation (PCAcc) method
Please
CLICK HERE to download the Matlab scripts.
PCAcc
method focuses on improving class separability by using the information
contained
in multiple PC subspaces to identify different types of cancers.
The
programs for the method are implemented in MATLAB. MATLAB 6.5 or higher
versions
should be installed for runing these programs.
The
package consists of 11 MATLAB scripts. Their functions are as follows:
1. class_AccPC.m: Main function of PCAcc
method
2. analgroup.m: dividing samples into two
groups
3. svdpca.m: applying PCA on the training
set and projecting the test set
onto PC subspaces
4. disprotpcsallpoints.m: exhibiting
samples distribution in PC plot
5. rotation.m: obtaining the resolutions
of all the PC subspaces and finding
the PC subspace with maximal resolution
6. rotpcs: maximizing the distance
between groups by using Fisher criterion
7. stats.m: selecting PC subspaces
8. dispsumR.m: exhibiting samples
distribution along the projected or the
accumulated PC axis
9. dispsumRs.m: calculating mean value,
standard deviation and resolution
10.
disclass.m: assigning the class label of a sample
11.
fastdist.m: calculating Euclidean distance between the samples with the
centroid of each group