Skip to main content
. 2020 Dec 24;21(1):57. doi: 10.3390/s21010057

Table 15.

Performance results obtained with 2 classes (crackle vs. other)—training: variable duration; testing: variable duration.

Classifiers Accuracy AUC F1Crack MCCCrack F1Other MCCOther
LDA_10MRMR 68.1 ± 0.2 74.7 ± 0.0 76.9 ± 0.1 49.4 ± 0.3 48.4 ± 0.5 33.3 ± 0.5
LDA_100MRMR 70.2 ± 0.3 76.3 ± 0.2 76.4 ± 0.2 52.2 ± 0.5 59.7 ± 1.6 42.7 ± 1.5
LDA_Full 68.5 ± 0.7 73.4 ± 1.1 74.9 ± 1.2 49.4 ± 1.1 57.5 ± 2.2 39.6 ± 2.0
SVMrbf_10MRMR 68.7 ± 0.2 72.2 ± 0.5 78.6 ± 0.1 52.2 ± 0.3 41.4 ± 1.0 31.7 ± 0.8
SVMrbf_100MRMR 72.6 ± 0.5 80.1 ± 0.8 78.6 ± 0.9 56.1 ± 0.9 61.8 ± 1.5 46.6 ± 0.9
SVMrbf_Full 71.2 ± 1.3 78.6 ± 1.4 77.2 ± 1.8 53.7 ± 2.0 60.6 ± 1.4 44.4 ± 1.3
RUSBoost_10MRMR 69.6 ± 0.3 76.0 ± 0.5 76.4 ± 0.6 51.2 ± 0.4 56.9 ± 2.3 40.1 ± 1.9
RUSBoost_100MRMR 71.0 ± 0.7 79.7 ± 0.4 76.9 ± 0.8 53.4 ± 1.1 61.0 ± 0.7 44.4 ± 1.0
RUSBoost_Full 69.9 ± 1.3 78.6 ± 0.9 75.0 ± 1.4 52.0 ± 2.0 62.4 ± 1.4 45.1 ± 2.1
CNN_dualInput 87.4 ± 1.4 84.9 ± 2.3 90.5 ± 0.9 73.0 ± 2.7 81.4 ± 3.0 73.0 ± 2.7
CNN_Spectrogram 86.5 ± 1.3 83.8 ± 2.3 89.8 ± 0.7 70.8 ± 2.5 79.9 ± 3.1 70.8 ± 2.5
CNN_melSpectrogram 85.1 ± 1.2 81.8 ± 2.0 88.9 ± 0.7 67.7 ± 2.6 77.4 ± 2.9 67.7 ± 2.6