On the Use of t-Distributed Stochastic Neighbor Embedding for Data Visualization and Classification of Individuals with Parkinson’s Disease

Table 2

Parameters and performance values of best scenarios according to its quality ratio (QR).

Task	DR methods	Parameter settings			Performance values
					QR (%)
Finger taps (T1)	PCA	−	−	−	79.82 ± 6.2
	Sammon’s	575	0.6	−	88.60 ± 8.6
	t-SNE	5000	0.6	27	99.42 ± 0.8
Finger to nose (T2)	PCA	−	−	−	86.26 ± 4.5
	Sammon’s	423	0.4	−	94.44 ± 7.8
	t-SNE	5000	0.4	5	99.71 ± 0.4
Pronation and supination (T3)	PCA	−	−	−	83.92 ± 15.2
	Sammon’s	742	0.6	−	93.57 ± 9.0
	t-SNE	1000	0.6	16	100 ± 0.0
Rest (T4)	PCA	−	−	−	75.44 ± 2.4
	Sammon’s	1080	0.4	−	86.26 ± 11.9
	t-SNE	3000	0.5	16	98.54 ± 2.0