Uncertain Johnson–Schumacher growth model with imprecise observations and k-fold cross-validation test

Regression is a powerful tool to study how the response variables vary due to changes of explanatory variables. Unlike traditional statistics or mathematics where data are assumed fairly accurate, we notice that the real-world data are messy and obscure; thus, the uncertainty theory seems more appropriate. In this paper, we focus on the residual analysis of the Johnson–Schumacher growth model, with parameter estimation performed by the least squares method, followed by the prediction intervals for new explanatory variables. We also propose a k-fold cross-validation method for model selection with imprecise observations. A numerical example illustrates that our approach will achieve better prediction accuracy.