Applying Hierarchical Bayesian Neural Network in Failure Time Prediction

Table 5

Sensitivity analysis.

# of hidden nodes	Training dataset	RMSE	MAD	MAPE	RMSPE
3	from 9 (mm) to 30 (mm)	0.40576	0.29048	1.108%	1.543%
	from 9 (mm) to 35 (mm)	0.38022	0.27297	1.081%	1.486%
	from 9 (mm) to 40 (mm)	0.37340	0.27121	1.058%	1.440%
4	from 9 (mm) to 30 (mm)	0.40568	0.29060	1.089%	1.558%
	from 9 (mm) to 35 (mm)	0.38070	0.27326	1.057%	1.487%
	from 9 (mm) to 40 (mm)	0.37366	0.27088	1.089%	1.467%
5	from 9 (mm) to 30 (mm)	0.40592	0.29036	1.082%	1.520%
	from 9 (mm) to 35 (mm)	0.38076	0.27350	1.056%	1.456%
	from 9 (mm) to 40 (mm)	0.37391	0.27065	1.062%	1.435%
6	from 9 (mm) to 30 (mm)	0.40606	0.29047	1.104%	1.536%
	from 9 (mm) to 35 (mm)	0.38051	0.27393	1.098%	1.489%
	from 9 (mm) to 40 (mm)	0.37357	0.27068	1.076%	1.461%