A new method for error degree estimation in numerical weather prediction via MKDA-based ordinal regression

EURASIP Journal on Advances in Signal Processing

Table 7 MAE-based quantitative evaluation of the proposed method and conventional methods

Target element	Ours	PM-WP	Ref [[13]]	SVM-EBC	SVOR-IMC
Temperature	0.4317	0.4458	0.3901 ^c	0.3840 ^b	0.3839 ^a
Relative humidity	0.7856^a	0.8017 ^b	0.8384 ^c	0.9030	0.9041
Wind velocity	0.8654	0.9035	0.8456 ^a	0.8645^c	0.8626^b

^aThe best method. ^bThe second-best method. ^cThe third-best method. PM-WP, proposed method without propagation. In addition, temperature, relative humidity, and wind velocity respectively represent temperature on the ground, relative humidity on the ground, and wind velocity on the ground. When the difference between the proposed method and the conventional method was confirmed by Welch’s t test, the values of the corresponding conventional method are italicized.