Skip to main content

Table 3 Comparison of classification performance of machine learning methods on validating and testing datasets

From: Diagnosis of thyroid nodules for ultrasonographic characteristics indicative of malignancy using random forest

Validating data Testing data
  AUC SEN F1 SPE PPV NPV   AUC SEN F1 SPE PPV NPV
RF 0.960 0.892 0.701 0.820 0.960 0.612   0.958 0.883 0.721 0.880 0.972 0.611
LR 0.965 0.875 0.720 0.900 0.977 0.600   0.965 0.875 0.720 0.900 0.977 0.600
SVM 0.952 0.862 0.713 0.920 0.981 0.582   0.962 0.896 0.740 0.900 0.977 0.643
NET 0.965 0.879 0.726 0.900 0.977 0.608   0.964 0.879 0.746 0.920 0.981 0.613
ELM 0.952 0.833 0.657 0.880 0.971 0.524   0.952 0.833 0.657 0.880 0.971 0.524
KNN 0.926 0.825 0.657 0.900 0.975 0.517   0.926 0.825 0.657 0.900 0.975 0.517
NB 0.940 0.871 0.672 0.820 0.959 0.569   0.940 0.871 0.672 0.820 0.959 0.569
ADAB 0.955 0.891 0.733 0.880 0.973 0.629   0.958 0.855 0.712 0.940 0.985 0.574
LOG 0.964 0.846 0.711 0.960 0.970 0.565   0.956 0.846 0.711 0.960 0.970 0.565
LDA 0.954 0.850 0.687 0.900 0.976 0.556   0.954 0.850 0.687 0.900 0.976 0.556
\