Application of Data Mining Algorithms for Dementia in People with HIV/AIDS

Table 3

Analysis of machine learning algorithms against the variation of HIV-associated dementia data.

		Logistic regression	Decision tree	Neural network	KNN	Random forest
Total features: 104
No PCA	Accuracy	0.7805	0.7317	0.5732	0.6098	0.8293
	Precision	0.7377	0.7451	0.5732	0.6415	0.7797
	Recall	0.9574	0.8085	1.0	0.7234	0.9787
Total features: 47 ()
90% of captured variation	Accuracy	0.7439	0.7073	0.5732	0.5976	0.8049
	Precision	0.7097	0.7091	0.5732	0.6296	0.7627
	Recall	0.9362	0.8298	1.0	0.7234	0.9574
Total features: 33
80% of captured variation	Accuracy	0.7195	0.7683	0.5610	0.5976	0.8049
	Precision	0.6935	0.7917	0.7619	0.6296	0.7541
	Recall	0.9149	0.8085	0.3404	0.7234	0.9787
Total features: 23
70% of captured variation	Accuracy	0.5610	0.6829	0.5854	0.5976	0.8659
	Precision	0.5679	0.6981	0.9333	0.6296	0.8462
	Recall	0.9787	0.7872	0.2979	0.7234	0.9362
Total features: 16
60% of captured variation	Accuracy	0.7317	0.7073	0.6463	0.5366	0.8049
	Precision	0.7049	0.7447	0.7500	0.5957	0.7818
	Recall	0.9149	0.7447	0.5745	0.5957	0.9149
Total features: 10
50% of captured variation	Accuracy	0.7195	0.6098	0.5610	0.5122	0.7317
	Precision	0.7000	0.6667	0.5696	0.5714	0.7358
	Recall	0.8936	0.6383	0.9574	0.5957	0.8298
Total features: 6
40% of captured variation	Accuracy	0.6220	0.5610	0.5366	0.5000	0.6829
	Precision	0.6111	0.6279	0.5570	0.5600	0.7059
	Recall	0.9362	0.5745	0.9362	0.5957	0.7660