-
Notifications
You must be signed in to change notification settings - Fork 17
/
robust_correlation.m
175 lines (146 loc) · 5.76 KB
/
robust_correlation.m
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
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
function results = robust_correlation(X,Y)
% General function that performs multiple tests on X and Y.
%
% FORMAT: results = robust_correlation(X,Y)
%
% INPUTS: X and Y are two vectors of the same length.
%
% OUTPUTS:
% result is a structure with one field for each output from the
% different tests performed.
% In particular the function:
%
% 1 - plots the data and generates univariate and bivariate histograms;
% 2 - performs the Henze-Zirkler test for normality (results.normality);
% 3 - computes conditional data and variances (results.conditional);
% 4 - tests for heteroscedasticity based on conditional variances
% (results.heteroscedasticity);
% 5 - checks for outliers (results.outliers);
% 6 - performs Pearson, Spearman, Bend and Skipped correlations.
%
% See also HZMVNTEST, CONDITIONAL, PEARSON, SPEARMAN, DETECT_OUTLIERS, BENDCORR, SKIPPED_CORRELATION.
%
% Cyril Pernet v1
% ---------------------------------
% Copyright (C) Corr_toolbox 2012
disp(' ');
disp('Correlation toolbox, Copyright (C) 2012. C. Pernet & G. Rousselet')
disp('This program comes with ABSOLUTELY NO WARRANTY')
disp('This is free software, and you are welcome to redistribute it')
disp('under certain conditions; see http://www.gnu.org/licenses/ for details.')
disp(' ');
%% data check
if size(X)~=size(Y)
error('X and Y must have the same size')
end
[r c] = size(X);
if r == 1 && c > 1
X = X';
Y = Y';
elseif r > 1 && c > 1
error('X and Y must be 2 vectors, more than 1 column/row detected')
end
level = 5/100;
%% 1 - plot the data and generate univariate and bivariate histograms
% figure 1 - histograms and scatter plot
corr_normplot(X,Y);
% figure 2 - Joint density
joint_density(X,Y,1)
%% 2 - performs the Henze- Zirkler test for normality
[results.normality.test_value,results.normality.p_value] = HZmvntest(X,Y,level);
%% 3 - get conditional data
[results.conditional.values,results.conditional.variances]=conditional(X,Y);
%% 4 - test for heteroscedasticity based on conditional variances
[h,results.heteroscedasticity.CI] = variance_homogeneity(X,Y,1);
if h == 0
results.heteroscedasticity.result = 'variance are equals';
else
results.heteroscedasticity.result = 'variances are inequals';
end
disp(' ')
fprintf('heteroscedasticity testing indicates that %s\n',results.heteroscedasticity.result)
fprintf('conditional variances = %g %g \n', results.conditional.variances)
%% 5 - check for outliers based on the MAD median rule & the IQR rule
results.outliers = detect_outliers(X,Y);
%% 6 - Pearson correlation
[r,t,pval,hboot,CI] = Pearson(X,Y);
results.Pearson.r = r;
results.Pearson.t = t;
results.Pearson.p = pval;
results.Pearson.CI = CI;
disp(' ')
if hboot == 1
fprintf('Pearson correlation is significant \n')
fprintf('r=%g p=%g CI=[%g %g] \n',r,pval,CI)
results.Pearson.result = 'Pearson correlation is significant';
else
fprintf('Pearson correlation is not significant \n')
fprintf('r=%g CI=[%g %g] \n',r,CI)
results.Pearson.result = 'Pearson correlation is not significant';
end
clear r t pval hboot CI
%% 6 - Bend correlation (remove effect of univariate outliers)
[r_bend,t_bend,p_bend,h,CI] = bendcorr(X,Y); % use the default 20% triming
results.bend_correlation.r = r_bend;
results.bend_correlation.t = t_bend;
results.bend_correlation.p = p_bend;
results.bend_correlation.CI = CI;
results.bend_correlation.h = h;
disp(' ')
if h == 1
fprintf('Bend correlation is significant \n')
fprintf('r=%g p=%g CI=[%g %g] \n',r_bend,p_bend,CI)
results.bend_correlation.result = 'Bend correlation is significant ';
else
fprintf('Bend correlation is not significant \n')
fprintf('r=%g CI=[%g %g] \n',r_bend,CI)
results.bend_correlation.result = 'Bend correlation is not significant ';
end
clear r_bend t_bend p_bend CI h
%% 7 - performs Spearman
[r,t,pval,hboot,CI] = Spearman(X,Y);
results.Spearman.r = r;
results.Spearman.t = t;
results.Spearman.p = pval;
results.Spearman.CI = CI;
disp(' ')
if hboot == 1
fprintf('Spearman correlation is significant \n')
fprintf('r=%g p=%g CI=[%g %g] \n',r,pval,CI)
results.Spearman.result = 'Spearman correlation is significant ';
else
fprintf('Spearman correlation is not significant \n')
fprintf('r=%g CI=[%g %g] \n',r,CI)
results.Spearman.result = 'Spearman correlation is not significant ';
end
clear r t pval hboot CI
%% 8 - skip correlation to remove effect of bivariate outliers
[r,t,h,outliers,hboot,CI]=skipped_correlation(X,Y);
results.Skipped_correlation.Pearson.r = r.Pearson;
results.Skipped_correlation.Spearman.r = r.Spearman;
results.Skipped_correlation.Pearson.t = t.Pearson;
results.Skipped_correlation.Spearman.t = t.Spearman;
results.Skipped_correlation.Pearson.CI = CI.Pearson;
results.Skipped_correlation.Spearman.CI = CI.Spearman;
results.Skipped_correlation.Pearson.h = h.Pearson;
results.Skipped_correlation.Spearman.h = h.Spearman;
disp(' ')
if hboot.Pearson == 1
fprintf('Pearson Skipped correlation is significant \n')
fprintf('r=%g CI=[%g %g] \n',r.Pearson,CI.Pearson)
results.Skipped_correlation.Pearson.result = 'Skipped correlation is significant ';
else
fprintf('Skipped correlation is not significant \n')
fprintf('r=%g CI=[%g %g]\n',r.Pearson,CI.Pearson)
results.Skipped_correlation.Pearson.result = 'Skipped correlation is not significant ';
end
if hboot.Spearman == 1
fprintf('Spearman Skipped correlation is significant \n')
fprintf('r=%g CI=[%g %g] \n',r.Spearman,CI.Spearman)
results.Skipped_correlation.Spearman.result = 'Skipped correlation is significant ';
else
fprintf('Spearman Skipped correlation is not significant \n')
fprintf('r=%g CI=[%g %g] \n',r.Spearman,CI.Spearman)
results.Skipped_correlation.Spearman.result = 'Skipped correlation is not significant ';
end
clear h r t outliers CI