Optimising the Selection of Input Variables to Increase the Predicting Accuracy of Shear Strength for Deep Beams

Table 5

A comparison between the efficiency of the proposed models.

Model	Distribution of data in terms of relative error for each model
	Training phase				Testing phase
	5%	10%	15%	20%	5%	10%	15%	20%
LWLR	62.78	87.78	95.56	98.89	31.96	60.04	75.00	85.87
RF	54.44	80.00	90.00	92.22	22.83	45.65	65.22	75.00
MLR	22.22	43.33	60.56	72.78	17.39	30.43	47.83	59.78
ELM	35.56	61.67	82.22	90.56	18.48	48.91	58.70	68.48