Table 2.

Mean Square Error and Bias for the ${\bar{y}}_{MBN}$ using Bump data set.

Fixed Bandwidth h = 0.2
	MSE	BIAS
Sample size
n = [150,150]	[0.0001628, 0.0001771]	[0.01275, 0.01330]
n = [200,200]	[0.0002220, 0.0002272]	[0.01489, 0.01507]
n = [250,250]	[0.0002555, 0.0002589]	[0.01598, 0.01609]
n = [300,300]	[0.0002718, 0.0002890]	[0.01648, 0.01700]
Fixed h = 0.5
n = [150,150]	[0.0001629, 0.0001773]	[0.01276, 0.01331]
n = [200,200]	[0.0002222, 0.0002274]	[0.01490, 0.01507]
n = [250,250]	[0.0002556, 0.0002591]	[0.01598, 0.01609]
n = [300,300]	[0.0002719, 0.0002893]	[0.01648, 0.01700]
${\mathit{h}}_{\mathit{w}\mathit{j}\mathbf{1}}$: Direct Plug-in Method, Wand and Jones [15]
n = [150,150]	[0.0001630, 0.0001765]	[0.01276, 0.01328]
n = [200,200]	[0.0002219, 0.0002265]	[0.01489, 0.01504]
n = [250,250]	[0.0002558, 0.0002583]	[0.01599, 0.01607]
n = [300,300]	[0.0002721, 0.0002885]	[0.01649, 0.01698]
${\mathit{h}}_{\mathit{w}\mathit{j}\mathbf{2}}$: Direct Plug-in Method, Wand and Jones [15]
n = [150,150]	[0.0001630, 0.0001766]	[0.01276, 0.01328]
n = [200,200]	[0.0002220, 0.0002268]	[0.01489, 0.01505]
n = [250,250]	[0.0002556, 0.0002584]	[0.01598, 0.01607]
n = [300,300]	[0.0002720, 0.0002885]	[0.01649, 0.01698]
${\mathit{h}}_{\mathit{D}\mathit{S}\mathbf{1}}$: Biased Cross-validation, Scott and Terrell [16]
n = [150,150]	[0.0001629, 0.0001773]	[0.01276, 0.01331]
n = [200,200]	[0.0002222, 0.0002275]	[0.01490, 0.01508]
n = [250,250]	[0.0002556, 0.0002592]	[0.01598, 0.01609]
n = [300,300]	[0.0002718, 0.0002893]	[0.01648, 0.01700]
${\mathit{h}}_{\mathit{D}\mathit{S}\mathbf{2}}$: Un-Biased Cross-validation, Scott and Terrell [16]
n = [150,150]	[0.0001628, 0.0001772]	[0.01275, 0.01331]
n = [200,200]	[0.0002220, 0.0002272]	[0.01489, 0.01507]
n = [250,250]	[0.0002555, 0.0002589]	[0.01598, 0.01609]
n = [300,300]	[0.0002717, 0.0002890]	[0.01648, 0.01700]