Table 4.

Wilcoxon rank sum test—pairwise comparisons of $MS{E}_{predict}^{\left(i\right)}$ and ${\overline{R}}^2{}_{predict}^{\left(i\right)}$.

		MSE ${}_{\mathit{predict}}^{\left(\mathit{i}\right)}$			${\overline{\mathit{R}}}^2{}_{\mathit{predict}}^{\left(\mathit{i}\right)}$
Filter	Pair	Z	Sig.	Best Model	Z	Sig.	Best Model
quartile = 1	log–exp	−4.94	0.000	Exponential	−5.89	0.000	Exponential
	pow–exp	−1.63	0.104	—	−2.97	0.003	Power
	pow–log	−6.92	0.000	Power	−7.05	0.000	Power
quartile = 2	log–exp	−4.32	0.000	Exponential	−5.45	0.000	Exponential
	pow–exp	−3.51	0.000	Power	−3.66	0.000	Power
	pow–log	−6.95	0.000	Power	−7.16	0.000	Power
quartile = 3	log–exp	−5.42	0.000	Exponential	−6.18	0.000	Exponential
	pow–exp	−3.11	0.002	Power	−3.93	0.000	Power
	pow–log	−8.10	0.000	Power	−7.94	0.000	Power