Table 27

Combining and filtering features for training RBF-epsilon regression SVM model on dataset2431 and on dataset579

train2431

train579

test579

test2431 10 × cross val

test2431

test579 10 × cross val





Method(s)

t-test

FN2431

R

MSE

R

MSE

FN579

R

MSE

R

MSE


1

0

84(84)

0.509

0.095

0.711

0.026

84(84)

0.484

0.054

0.562

0.079

1

2

45(43.9)

0.494

0.100

0.687

0.027

22(21)

0.467

0.058

0.449

0.091

2

0

23(23)

0.379

0.105

0.640

0.029

23(23)

0.372

0.065

0.500

0.087

2

2

19(19)

0.363

0.113

0.641

0.029

11(11.1)

0.330

0.061

0.548

0.082

11

0

1360(1360)

0.246

0.109

0.559

0.033

1360(1360)

0.192

0.055

0.469

0.088

11

2

424(394.3)

0.576

0.070

0.471

0.036

195(179.3)

0.467

0.032

0.431

0.091

1,2

0

107(107)

0.450

0.104

0.701

0.026

107(107)

0.493

0.083

0.533

0.083

1,2

2

64(62.9)

0.444

0.105

0.704

0.026

33(32.1)

0.466

0.069

0.537

0.083

2,11

0

1383(1383)

0.491

0.101

0.721

0.025

1383(1383)

0.521

0.067

0.669

0.068

2,11

2

443(413.3)

0.480

0.094

0.688

0.027

206(190.4)

0.489

0.051

0.641

0.071

1,11

0

1444(1444)

0.492

0.096

0.784

0.022

1444(1444)

0.423

0.051

0.646

0.070

1,11

2

469(438.2)

0.485

0.102

0.765

0.023

217(200.3)

0.431

0.062

0.556

0.080

1,2,11

0

1467(1467)

0.518

0.098

0.767

0.023

1467(1467)

0.546

0.064

0.662

0.070

1,2,11

2

488(457.2)

0.528

0.091

0.750

0.023

228(211.4)

0.521

0.058

0.640

0.071

1,2,4,5,11

0

1523(1523)

0.523

0.099

0.760

0.023

1523(1523)

0.548

0.067

0.643

0.070

1,2,4,5,11

2

522(489.5)

0.526

0.092

0.746

0.024

244(226.8)

0.516

0.061

0.639

0.071

1,2,4,5,11,13

0

1566(1566)

0.523

0.099

0.710

0.027

1566(1566)

0.549

0.067

0.613

0.076

1,2,4,5,11,13

2

535(500.9)

0.526

0.092

0.728

0.025

249(232.7)

0.517

0.061

0.628

0.072


Method numbers are from Table 1.

Models trained on dataset2431 and testing performed with dataset579 and 10 × cross validation on dataset2431, features removed by increasing stringency of t-test of individual feature to activity from dataset2431.

Models trained on dataset579 and testing performed with dataset2431 and 10 × cross validation on dataset579, features removed by increasing stringency of t-test of individual feature to activity from dataset579.

Feature numbers in parentheses are the average number of features in cross validations.

Entries in bold are column maximums from their respective tables and are provided as visual indicators. Italicized entries are column maximums within the table and are again provided as visual indicators.

Peek BMC Bioinformatics 2007 8:182   doi:10.1186/1471-2105-8-182

Open Data