Dysphonic Voice Pattern Analysis of Patients in Parkinson’s Disease Using Minimum Interclass Probability Risk Feature Selection and Bagging Ensemble Learning Methods

Table 5

Feature selection by means of the interclass probability risk (ICPR) and modified Kullback-Leibler divergence (MKLD) methods. Bold values are selected features for pattern classifications (ICPR 0.6; MKLD 1).

Features	ICPR	MKLD
MDVP:F0	0.59	1.02
MDVP:Fhi	0.71	0.26
MDVP:Flo	0.67	1.13
HNR	0.66	0.73
RPDE	0.71	0.25
D2	0.76	0.22
Spread1	0.56	0.89
Spread2	0.69	0.46
MDVP-LDA	0.59	1.96
Shimmer-LDA	0.46	5.72
Nonlinear-LDA	0.43	2.17