1 %> @brief Paired Vector Comparer base
class
3 %> Compares two vectors. Number of elements must be the same in both vector. This is applied to:
4 %> @arg comparing two classifiers (see @ref reptt_sgs, which has a @ref reptt_sgs::vectorcomp property)
5 %> @arg measusing the diversity of two classifiers (example: @ref reptt_xornorm)
7 %> Depending on the
case (i.e., the @ref
vectorcomp descendant considered), the vectors must be classification rates or the
class predictions
8 %> themselves. See the descendant classes (i.e., classes vectorcomp_*)
10 methods(Access=protected)
12 function z = do_test(o, vv)
18 o.classtitle = 'Vector Comparer';
19 %o.color = [118, 176, 189]/255;
20 o.color = [244, 192, 34]/255;
25 % May return one grade or a vector thereof
26 function z = test(o, v1, v2)
27 z = o.do_test(v1, v2);
30 %> Cross-test of many vectors
32 %> Performs the test with all combinations of the columns of R
34 %> The resulting table should be read
as: the row represents the first variable
35 %> and the column represents the second variable.
38 %> @arg for a 2-sided t-test: H0: mean_row_i - mean_col_j = 0
39 %> @arg for a right-tail t-test: H0: mean_row_i - mean_col_j < 0
42 %> @param R a matrix of shape [v1, v2, v3, v4, ...]
43 %> @return a matrix [number of v's]x[number of v's]
44 function M = crosstest(o, R)
50 meas = o.test(R(:, i), R(:, j));
Paired Vector Comparer base class.
Analysis Session (AS) base class.