TABLE 5.

Neural Network Analysis model with NTanH(10) summary of training and validation model and variable importance assuming dependent resampled inputs for the prediction of response variable Edamame Harvest Quality Index (EHQI) based on phenological (Ve, emergence; R1, first flower) day-of-year (DOY), and thermal functions (GDD, growing degree days) for soybean cultivar R09-345.

Model summary	Training	Validation
R2	0.668	0.783
RMSE	0.017	0.020
Variable importance	Main effect	Total effect
#Days Ve-R1	0.069	0.201
HarvestDate_DOY	0.155	0.155
#Days Ve-Harvest	0.135	0.135
#Days R1-Harvest	0.132	0.132
GDD Ve-Harvest	0.082	0.109
GDD R1-Harvest	0.103	0.103
R1Date_DOY	0.100	0.100
PlantingDate_DOY	0.082	0.082
VeDate_DOY	0.078	0.078
GDD Ve-R1	0.064	0.064