Skip to main content

Table 3 Classification performances (mean accuracy, specificity, sensitivity, and MCC values with their standard deviations) computed on 10 replicates of 5-fold cross validation applied to the training dataset

From: A filter approach for feature selection in classification: application to automatic atrial fibrillation detection in electrocardiogram recordings

Feature selection (Training dataset)

Classification performances (Training dataset)

Dim

Added feature

KI

Accuracy

Specificity

Sensitivity

MCC

\(\gamma\)-metric

1

\(sd_2\)

1.000

\(0.9993\ (0.000228)\)

\(0.9997\ (0.000172)\)

\(0.9933\ (0.000030)\)

\(0.994\ (0.0021)\)

2

\(sd_3\)

1.000

\(0.9996\ (0.000164)\)

\(0.9997\ (0.000127)\)

\(0.9963\ (0.000024)\)

\(0.997\ (0.0015)\)

3

\(sd_1\)

1.000

\(0.9996\ (0.000181)\)

\(0.9998\ (0.000134)\)

\(0.9963\ (0.000025)\)

\(0.997\ (0.0017)\)

4

pNN50

0.971

\(0.9997\ (0.000174)\)

\(0.9999\ (0.000139)\)

\(0.9962\ (0.000019)\)

\(0.997\ (0.0016)\)

5

\(sd_4\)

1.000

\(0.9996\ (0.000197)\)

\(0.9999\ (0.000104)\)

\(0.9960\ (0.000026)\)

\(0.997\ (0.0018)\)

6

RMSSD

0.926

\(0.9998\ (0.000133)\)

\(0.9999\ (0.000080)\)

\(0.9973\ (0.000018)\)

\(0.998\ (0.0012)\)

7

SDSD

0.918

\(0.9998\ (0.000143)\)

\(0.9999\ (0.000093)\)

\(0.9972\ (0.000017)\)

\(0.998\ (0.0013)\)

8

\(m_2\)

1.000

\(0.9997\ (0.000145)\)

\(0.9999\ (0.000085)\)

\(0.9969\ (0.000023)\)

\(0.998\ (0.0013)\)

9

\(m_3\)

1.000

\(0.9997\ (0.000145)\)

\(0.9999\ (0.000086)\)

\(0.9968\ (0.000021)\)

\(0.997\ (0.0013)\)

10

\(m_1\)

1.000

\(0.9997\ (0.000151)\)

\(0.9999\ (0.000100)\)

\(0.9966\ (0.000026)\)

\(0.997\ (0.0014)\)

11

SDNNIDX

0.981

\(0.9998\ (0.000132)\)

\(1.0000\ (0.000065)\)

\(0.9971\ (0.000021)\)

\(0.998\ (0.0012)\)

12

\(sd_5\)

0.995

\(0.9998\ (0.000152)\)

\(1.0000\ (0.000065)\)

\(0.9971\ (0.000024)\)

\(0.998\ (0.0014)\)

13

\(m_4\)

1.000

\(0.9998\ (0.000154)\)

\(1.0000\ (0.000061)\)

\(0.9974\ (0.000022)\)

\(0.998\ (0.0014)\)

14

MADRR

0.990

\(0.9998\ (0.000136)\)

\(0.9999\ (0.000072)\)

\(0.9973\ (0.000021)\)

\(0.998\ (0.0012)\)

15

\(sd_6\)

0.967

\(0.9998\ (0.000132)\)

\(1.0000\ (0.000065)\)

\(0.9973\ (0.000021)\)

\(0.998\ (0.0012)\)

16

\(m_5\)

1.000

\(0.9998\ (0.000134)\)

\(0.9999\ (0.000091)\)

\(0.9973\ (0.000018)\)

\(0.998\ (0.0012)\)

Mean decrease accuracy

1

\(sd_3\)

1.000

\(0.9995\ (0.000209)\)

\(0.9998\ (0.000151)\)

\(0.9955\ (0.002087)\)

\(0.996\ (0.0019)\)

2

\(sd_2\)

0.876

\(0.9996\ (0.000162)\)

\(0.9998\ (0.000136)\)

\(0.9962\ (0.002022)\)

\(0.996\ (0.0015)\)

3

\(sd_4\)

0.902

\(0.9996\ (0.000175)\)

\(0.9998\ (0.000127)\)

\(0.9963\ (0.002363)\)

\(0.997\ (0.0016)\)

4

\(m_3\)

0.852

\(0.9996\ (0.000220)\)

\(0.9998\ (0.000144)\)

\(0.9963\ (0.003057)\)

\(0.997\ (0.0020)\)

5

\(sd_5\)

0.868

\(0.9996\ (0.000179)\)

\(0.9999\ (0.000130)\)

\(0.9963\ (0.002221)\)

\(0.997\ (0.0016)\)

6

\(m_2\)

0.886

\(0.9996\ (0.000151)\)

\(0.9998\ (0.000114)\)

\(0.9962\ (0.002100)\)

\(0.997\ (0.0014)\)

7

\(sd_6\)

0.883

\(0.9997\ (0.000171)\)

\(0.9999\ (0.000114)\)

\(0.9963\ (0.002735)\)

\(0.997\ (0.0016)\)

8

\(m_4\)

0.858

\(0.9997\ (0.000122)\)

\(0.9999\ (0.000100)\)

\(0.9966\ (0.001843)\)

\(0.997\ (0.0011)\)

9

\(sd_7\)

0.838

\(0.9997\ (0.000161)\)

\(0.9999\ (0.000094)\)

\(0.9966\ (0.002583)\)

\(0.997\ (0.0015)\)

10

\(sd_1\)

0.815

\(0.9997\ (0.000177)\)

\(0.9999\ (0.000120)\)

\(0.9966\ (0.002512)\)

\(0.997\ (0.0016)\)

11

\(sd_8\)

0.802

\(0.9997\ (0.000171)\)

\(0.9999\ (0.000109)\)

\(0.9965\ (0.002426)\)

\(0.997\ (0.0016)\)

12

\(sd_9\)

0.796

\(0.9997\ (0.000161)\)

\(0.9999\ (0.000081)\)

\(0.9965\ (0.002487)\)

\(0.997\ (0.0015)\)

13

\(m_1\)

0.801

\(0.9997\ (0.000156)\)

\(0.9999\ (0.000119)\)

\(0.9965\ (0.002089)\)

\(0.997\ (0.0014)\)

14

\(sd_{10}\)

0.824

\(0.9997\ (0.000191)\)

\(0.9999\ (0.000102)\)

\(0.9965\ (0.002697)\)

\(0.997\ (0.0018)\)

15

\(m_5\)

0.851

\(0.9997\ (0.000144)\)

\(0.9999\ (0.000090)\)

\(0.9965\ (0.002118)\)

\(0.998\ (0.0013)\)

16

SDSD

0.853

\(0.9998\ (0.000151)\)

\(0.9999\ (0.000070)\)

\(0.9971\ (0.002154)\)

\(0.998\ (0.0014)\)

Mean decrease gini

1

\(sd_3\)

0.634

\(0.9996\ (0.000210)\)

\(0.9998\ (0.000128)\)

\(0.9955\ (0.002808)\)

\(0.996\ (0.0019)\)

2

\(sd_2\)

0.778

\(0.9996\ (0.000176)\)

\(0.9998\ (0.000112)\)

\(0.9964\ (0.002226)\)

\(0.997\ (0.0016)\)

3

\(m_3\)

0.790

\(0.9996\ (0.000157)\)

\(0.9998\ (0.000101)\)

\(0.9964\ (0.002343)\)

\(0.997\ (0.0014)\)

4

\(m_4\)

0.883

\(0.9997\ (0.000154)\)

\(0.9998\ (0.000118)\)

\(0.9966\ (0.002061)\)

\(0.997\ (0.0014)\)

5

\(sd_4\)

0.886

\(0.9997\ (0.000164)\)

\(0.9998\ (0.000147)\)

\(0.9966\ (0.001517)\)

\(0.997\ (0.0015)\)

6

\(m_2\)

0.911

\(0.9997\ (0.000173)\)

\(0.9998\ (0.000136)\)

\(0.9966\ (0.002243)\)

\(0.997\ (0.0016)\)

7

\(m_5\)

0.910

\(0.9997\ (0.000206)\)

\(0.9999\ (0.000113)\)

\(0.9964\ (0.002639)\)

\(0.997\ (0.0019)\)

8

\(sd_5\)

0.915

\(0.9997\ (0.000135)\)

\(0.9999\ (0.000106)\)

\(0.9964\ (0.002013)\)

\(0.997\ (0.0012)\)

9

\(sd_1\)

0.914

\(0.9997\ (0.000168)\)

\(0.9999\ (0.000113)\)

\(0.9965\ (0.002046)\)

\(0.997\ (0.0015)\)

10

\(sd_6\)

0.929

\(0.9997\ (0.000162)\)

\(0.9999\ (0.000115)\)

\(0.9966\ (0.002147)\)

\(0.997\ (0.0015)\)

11

\(m_1\)

0.931

\(0.9997\ (0.000176)\)

\(0.9999\ (0.000109)\)

\(0.9962\ (0.002669)\)

\(0.997\ (0.0016)\)

12

\(sd_7\)

0.912

\(0.9997\ (0.000151)\)

\(0.9999\ (0.000108)\)

\(0.9965\ (0.002256)\)

\(0.997\ (0.0014)\)

13

\(m_6\)

0.919

\(0.9997\ (0.000128)\)

\(0.9999\ (0.000091)\)

\(0.9963\ (0.001798)\)

\(0.997\ (0.0012)\)

14

\(sd_8\)

0.922

\(0.9997\ (0.000170)\)

\(0.9999\ (0.000085)\)

\(0.9963\ (0.002337)\)

\(0.997\ (0.0016)\)

15

\(sd_9\)

0.912

\(0.9997\ (0.000156)\)

\(0.9999\ (0.000082)\)

\(0.9962\ (0.002653)\)

\(0.997\ (0.0014)\)

16

\(sd_{10}\)

0.880

\(0.9997\ (0.000143)\)

\(0.9999\ (0.000102)\)

\(0.9962\ (0.002302)\)

\(0.997\ (0.0013)\)

AUC

1

\(sd_3\)

1.000

\(0.9995\ (0.000193)\)

\(0.9998\ (0.000149)\)

\(0.9955\ (0.002061)\)

\(0.996\ (0.0018)\)

2

\(sd_2\)

0.861

\(0.9996\ (0.000163)\)

\(0.9998\ (0.000114)\)

\(0.9964\ (0.002626)\)

\(0.997\ (0.0015)\)

3

\(m_3\)

0.802

\(0.9996\ (0.000182)\)

\(0.9998\ (0.000132)\)

\(0.9963\ (0.002337)\)

\(0.997\ (0.0017)\)

4

\(sd_4\)

0.857

\(0.9996\ (0.000184)\)

\(0.9998\ (0.000122)\)

\(0.9963\ (0.002353)\)

\(0.997\ (0.0017)\)

5

\(m_2\)

0.879

\(0.9996\ (0.000191)\)

\(0.9998\ (0.000109)\)

\(0.9962\ (0.002470)\)

\(0.997\ (0.0018)\)

6

\(sd_5\)

0.897

\(0.9996\ (0.000165)\)

\(0.9998\ (0.000110)\)

\(0.9963\ (0.002512)\)

\(0.997\ (0.0015)\)

7

\(m_4\)

0.924

\(0.9997\ (0.000150)\)

\(0.9998\ (0.000129)\)

\(0.9966\ (0.002119)\)

\(0.997\ (0.0014)\)

8

\(sd_1\)

0.970

\(0.9997\ (0.000172)\)

\(0.9998\ (0.000136)\)

\(0.9966\ (0.002388)\)

\(0.997\ (0.0016)\)

9

\(sd_6\)

0.978

\(0.9997\ (0.000115)\)

\(0.9999\ (0.000093)\)

\(0.9966\ (0.002001)\)

\(0.997\ (0.0011)\)

10

\(m_1\)

0.907

\(0.9997\ (0.000186)\)

\(0.9999\ (0.000097)\)

\(0.9965\ (0.002535)\)

\(0.997\ (0.0017)\)

11

\(sd_7\)

0.924

\(0.9997\ (0.000149)\)

\(0.9999\ (0.000087)\)

\(0.9966\ (0.001939)\)

\(0.997\ (0.0014)\)

12

\(m_5\)

0.941

\(0.9997\ (0.000154)\)

\(0.9999\ (0.000098)\)

\(0.9964\ (0.002420)\)

\(0.997\ (0.0014)\)

13

\(sd_8\)

0.995

\(0.9997\ (0.000138)\)

\(0.9999\ (0.000082)\)

\(0.9964\ (0.002398)\)

\(0.998\ (0.0013)\)

14

\(sd_9\)

1.000

\(0.9997\ (0.000135)\)

\(0.9999\ (0.000087)\)

\(0.9964\ (0.001707)\)

\(0.997\ (0.0012)\)

15

\(sd_{10}\)

0.947

\(0.9997\ (0.000159)\)

\(0.9999\ (0.000094)\)

\(0.9964\ (0.002237)\)

\(0.997\ (0.0015)\)

16

RMSSD

0.951

\(0.9998\ (0.000162)\)

\(0.9999\ (0.000094)\)

\(0.9970\ (0.002176)\)

\(0.998\ (0.0015)\)

  1. Results are presented according to the number of features (from 1 to 16) and to the KI calculated on 150 replications of the ranking given by the four feature selection approaches (\(\gamma\)-metric, MDA, MDG, and AUC)