Fig. 3
From: Classification of vasovagal syncope from physiological signals on tilt table testing
The partial dependence plot by D1(KNN imputation) data. According to feature importance, three tilting features named "CV_SBPV" (coefficient of variance of systolic BP variability), "CV_DBPV" (coefficient of variance of diastolic BP variability), and "LFnu_SBPV" (Low-frequency normalized power of systolic BP variability) were chosen, and this feature combination produced the best accuracy. Here, x axis CV_SBPV_T, CV_DBPV_T and Lfnu_SBPV_T represent the fixed values of target features and y axis represents the probability of predicted risk of a positive test. The subjects were more likely to test positive with higher CV_SBPV_T and CV_DBPV_T, while positive tests had lower Lfnu_SBPV_T