Multisensory Design of Electric Shavers Based on Kansei Engineering and Artificial Neural Networks

Table 6

Comparison and analysis results of prediction models.

Sample	Adj		Traditional/fashion-forward	Soft/harder-edged	Cheap/premium	Sophisticated/minimalistic
Test 1	QTTI	AER	0.49	0.14	0.17	0.29
		RER	10.18%	2.90%	3.39%	6.15%
	BPNN	AER	0.60	1.05	0.30	0.47
		RER	12.51%	21.28%	6.12%	9.99%
	GA-BPNN	AER	0.38	0.18	0.37	0.32
		RER	7.88%	3.62%	7.45%	6.79%
Test 2	QTTI	AER	0.35	0.19	0.25	0.38
		RER	8.94%	5.89%	6.51%	8.42%
	BPNN	AER	0.56	0.43	0.41	0.46
		RER	14.24%	13.28%	10.53%	10.30%
	GA-BPNN	AER	0.70	0.28	0.58	0.33
		RER	17.70%	8.65%	14.91%	7.25%
Test 3	QTTI	AER	0.16	0.21	0.41	0.12
		RER	4.74%	4.52%	13.43%	3.73%
	BPNN	AER	2.37	0.12	0.79	0.46
		RER	66.92%	2.59%	26.65%	14.09%
	GA-BPNN	AER	0.30	0.40	0.23	0.21
		RER	8.38%	8.65%	7.57%	6.32%
Test 4	QTTI	AER	0.67	0.37	0.20	0.25
		RER	22.87%	12.96%	8.08%	5.75%
	BPNN	AER	0.43	1.04	1.42	0.32
		RER	11.76%	36.25%	56.59%	7.43%
	GA-BPNN	AER	0.57	0.43	0.28	0.37
		RER	15.71%	15.00%	11.17%	8.39%
AER (average)	QTTI:		11.67%	6.57%	9.82%	6.01%
	BPNN:		26.35%	18.35%	24.97%	10.54%
	GA-BPNN:		12.41%	8.98%	10.27%	7.18%

Notes: RER, relative error rate; AER, average error rate.