[Master Index]
[Index for export/brede]
brede_loc_test_equal
(export/brede/brede_loc_test_equal.m)
Function Synopsis
R = brede_loc_test_equal(L1, L2, varargin);
Help text
brede_loc_test_equal - Test if 'loc' structures are equal
function R = brede_loc_test_equal(L1, L2, varargin);
Input: L1 First 'loc' structure or matrix
L2 Second 'loc' structure or matrix
Property: Type [ {Means} | Covars | EuclideanPermutation
MahalanobisPermutation |
MedianPermutation | PeelingPermutation ]
Dimensions [ {xyz} | x | y | z | xy | xz | ... ]
Restrict the tested dimensions
Permutations [ {10000} | positive integer ]
Output: R P-value
Test whether the distribution of the coordinates in L1 and L2
are the same.
The test for 'Type'='Means' is the Hotelling's T2 test, that
tests for the equality of means and assuming Gaussian
distributions with equal (homogeneous) covariances.
When 'Type'='Covars' then the equality for homogeneous covariance
is tested.
For 'Type'='MahalanobisPermutation' a permutation test is
employed on the Mahalanobis distance between the two location
sets. This should be the same as a permutation Hotelling T2
test. The number of permutations is controlled by the
'permutation' property. The p-value is returned.
'Type'='MedianPermutation' is permutation test on the sum over
dimensions of the absolute value of the difference between the
medians of the two groups. The median is calculated separately
for each dimension.
Ref: Mardia, Multivariate Analysis (1979) pp. 77, 139, 334.
Arnold et al, Blood 89(3):823+ (1997)
See also BREDE, BREDE_LOC, BREDE_LOC_LOC2MAT, BREDE_LOC_MEAN,
BREDE_LOC_MEDIAN, BREDE_LOC_TEST_GAUSSIAN,
BREDE_CDF_F, BREDE_MAT_MEAN.
$Id: brede_loc_test_equal.m,v 1.10 2005/06/04 01:23:24 fnielsen Exp $
Cross-Reference Information
This function calls
Produced by mat2html on Fri Jun 20 17:43:05 2008
Cross-Directory links are: OFF