International Journal of Academic and Applied Research (IJAAR)   Year: 2019 | Volume: 3 | Issue: 3 | Page No.: 8-11 Ability of Covariance Matrix in Bi-Response Multi-Prredictor Penalized Spline Model Through Longitudinal Data Simulation Anna Islamiyati, Fatmawati, Nur Chamidah Abstract: : Bi-response longitudinal data is assumed to have a correlation between responses and observations on the same subject. This causes a correlation between errors. To overcome this problem, we can use a penalized spline model that involves weighting. In this study, the weight used is the covariance matrix. Based on longitudinal data simulations that contain two responses and three predictors, a small GCV value is obtained from the penalized spline model that involves the covariance matrix. This value is compared with the penalized spline model without the covariance matrix. This shows that we need to involve a covariance matrix in estimating bi-response multi-predictor longitudinal data with a penalized spline model.