Online Program

Smooth Ridge Model for Computer Experiments (309844)

Keywords: Smooth model, interpolation, ridge regression

In Statistics, smoothness of a data set bears great importance when the purpose is to approximate the underlying function in an attempt to capture the meaningful patterns in data while removing noise and random phenomena. In this respect, smooth interpolation is a powerful concept in the modern age of data science and machine learning. In this study, a Smooth Ridge (SR) model is proposed that has found motivation from an advanced adaptation of literature on Smooth supersaturated models. The proposed SR model provides a new direction in the realm of smooth interpolation for computer experiments. It is shown that ridge regression and standard regression models are special cases of SR model which makes it more flexible. Moreover, the model bears inherent features of smoothness along with efficient interpolation, even in the presence of a singular design model matrix. Theoretical results provide sufficient conditions to prove the improved performance of SR model supported by simulation studies and real life COVID-19 data. These studies show that proposed model outperforms its contemporary models in terms of efficient predictions.