pwcmp/pw_plot_cmp_table.m at master · mantiuk/pwcmp · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
function pw_plot_cmp_table( T, group_col, condition_cols, observer_col, selection_col )
arguments
    T table
    group_col char
    condition_cols (1,2) cell = { 'condition_A', 'condition_B' }
    observer_col char = 'observer'
    selection_col char = 'selected_A'
end
% Plot a table with pairwise comparisons.
%
% pw_plot_cmp_table( T, group_col, condition_cols, observer_col, selection_col )
%
% T is the table with the results
% group_col - name of the column used to group the results. For example, if
%             the scaling is performed separately for each scene/content,
%             this should be the column with the ID of the content. Pass
%             an empty array ([]) if there are no groups in the table.
% conditions_col - a cell array with the name of the column string two
%             compared conditions, for example { 'condition_A', 'condition_B' }
% observer_col - the name of the column storing the ids of the observers
% selection_col - the name of the variable indicating whether the first
%             condition was selected. The values must be 0 or 1

if isempty(group_col)
    GRs = {};
    N_g = 1;
else
    GRs = unique( T.(group_col) ); % list of groups
    N_g = length( GRs );
end
C = unique( cat( 1, T.(condition_cols{1}), T.(condition_cols{2}) ) ); % all conditions

% The observer column must contain strings
if isnumeric(T.(observer_col))
    T.(observer_col) = mat2cell(num2str(T.(observer_col)), ones(height(T),1));
end

N = length(C);


% for each group
for gg=1:N_g

    if isempty(group_col)
        Ds = T;
        group = 'all';
    else
        Ds = T( strcmp( T.(group_col), GRs{gg} ), :);
        group = GRs{gg};
    end

    fprintf( 1, 'Group: %s\n', group );

    OBSs = unique( Ds.(observer_col) );

    MM = zeros(length(OBSs), N*N);

    for oo=1:length( OBSs )

        Dso = Ds(strcmp( Ds.(observer_col), OBSs{oo} ), :);

        M = zeros(N,N);

        for kk=1:height(Dso)

            % Find the indexes of both conditions
            c1 = find( strcmp( Dso.(condition_cols{1})(kk), C ), 1 );
            c2 = find( strcmp( Dso.(condition_cols{2})(kk), C ), 1 );

            if( isempty( c1 ) )
                error( 'Cannot find condition %s', Dso.(condition_cols{1})(kk) );
            end
            if( isempty( c2 ) )
                error( 'Cannot find condition %s', Dso.(condition_cols{2})(kk) );
            end

            if( Dso.(selection_col)(kk) )
                M(c1,c2) = M(c1,c2) + 1; % c1 better than c2
            else
                M(c2,c1) = M(c2,c1) + 1; % c2 better than c1
            end
        end

        MM(oo,:) = M(:);

    end

    cmp_table = reshape( sum(MM), N, N );

    figure(gg);
    clf;
    C_noun = strrep( C, '_', ' ' );
    heatmap( C_noun, C_noun, cmp_table );
    title( 'The condition in a row selected over the condition in a column' );

end

end