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
|
#include <BALLTestConfig.h>
#include <BALL/CONCEPT/classTest.h>
#include <BALL/QSAR/QSARData.h>
#include <BALL/QSAR/pcrModel.h>
#include <BALL/QSAR/featureSelection.h>
using namespace BALL;
using namespace BALL::QSAR;
START_TEST(Regression)
PRECISION(1E-5)
QSARData data;
std::multiset<int> activities;
activities.insert(0);
data.readSDFile(BALL_TEST_DATA_PATH(QSAR_test.sdf),activities,0,0);
data.centerData(true);
PCRModel pcr(data,0.95);
double cutoff = 0.001;
CHECK(RemoveColinearFeatures)
FeatureSelection fs(pcr);
fs.setQualityIncreaseCutoff(cutoff);
double cor_threshold=0.95;
fs.removeHighlyCorrelatedFeatures(cor_threshold);
TEST_EQUAL(pcr.getDescriptorIDs()->size(),26);
pcr.validation->crossValidation(5);
double q2 = pcr.validation->getQ2();
TEST_REAL_EQUAL(q2,0.653379);
pcr.readTrainingData();
pcr.train();
RESULT
CHECK(BackwardSelection)
PCRModel pcr1(data);
pcr1 = pcr; // copy model
FeatureSelection fs(pcr1);
fs.setQualityIncreaseCutoff(cutoff);
fs.backwardSelection(5);
TEST_EQUAL(pcr1.getDescriptorIDs()->size(),20);
pcr1.validation->crossValidation(5);
double q2 = pcr1.validation->getQ2();
TEST_REAL_EQUAL(q2,0.718207);
RESULT
CHECK(StepwiseSelection)
PCRModel pcr1(data,0.95);
pcr1 = pcr; // copy model
FeatureSelection fs(pcr1);
fs.setQualityIncreaseCutoff(cutoff);
fs.stepwiseSelection(5);
TEST_EQUAL(pcr1.getDescriptorIDs()->size(),5);
pcr1.validation->crossValidation(5);
double q2 = pcr1.validation->getQ2();
TEST_REAL_EQUAL(q2,0.69725);
RESULT
CHECK(ForwardSelection)
PCRModel pcr1(data,0.95);
pcr1 = pcr; // copy model
FeatureSelection fs(pcr1);
fs.setQualityIncreaseCutoff(cutoff);
fs.forwardSelection(5);
TEST_EQUAL(pcr1.getDescriptorIDs()->size(),5);
pcr1.validation->crossValidation(5);
double q2 = pcr1.validation->getQ2();
TEST_REAL_EQUAL(q2,0.69725);
RESULT
CHECK(RemoveLowResponseCorrelation)
PCRModel pcr1(data,0.95);
pcr1 = pcr; // copy model
FeatureSelection fs(pcr1);
fs.setQualityIncreaseCutoff(cutoff);
double cor_threshold=0.15;
fs.removeLowResponseCorrelation(cor_threshold);
TEST_EQUAL(pcr1.getDescriptorIDs()->size(),25);
pcr1.validation->crossValidation(5);
double q2 = pcr1.validation->getQ2();
TEST_REAL_EQUAL(q2,0.635891);
RESULT
CHECK(TwinScan)
PCRModel pcr1(data,0.95);
pcr1 = pcr; // copy model
FeatureSelection fs(pcr1);
fs.setQualityIncreaseCutoff(cutoff);
fs.twinScan(5);
TEST_EQUAL(pcr1.getDescriptorIDs()->size(), 10);
pcr1.validation->crossValidation(5);
double q2 = pcr1.validation->getQ2();
TEST_REAL_EQUAL(q2, 0.672368);
RESULT
END_TEST
|
