Skip to main content
. 2023 Jul 20;13(14):2417. doi: 10.3390/diagnostics13142417

Table 26.

Best results for classifiers without using feature selection (kNN*) and after using feature selection (BPSO-kNN*) in terms of accuracy, number of features, and improvement rate.

Dataset KNN* BPSO-KNN Improvement Rate
Accuracy No. Features Accuracy No. Features Features Reduction Accuracy
All 0.7409 31 0.7628 18 41.94% 2.19%
Caucasion 0.7483 30 0.8080 13 56.67% 5.96%
Hispanic 0.7724 30 0.7968 14 53.33% 2.44%
Females 0.7373 30 0.8136 13 56.67% 7.63%
Males 0.6987 30 0.7885 11 63.33% 8.97%
Age ≤ 50 0.7431 31 0.8624 15 51.61% 11.93%
Age > 50 0.7333 31 0.8061 17 45.16% 7.27%