CFE 2019: Start Registration
View Submission - CMStatistics
Title: Non-crossing p-splines quantile regression on time varying coefficient model of crude palm oil production in Indonesia Authors:  Yudhie Andriyana - Universitas Padjadjaran (Indonesia) [presenting]
Abstract: The use of nonparametric approaches in a quantile regression technique is important, because some parts of quantile levels are very hard to specify parametrically. Unfortunately, similar to the parametric approach, the estimated regression quantile curves using nonparametric techniques often cross each other, which can be very annoying for interpretations and further analysis. We are concerned with a non-crossing p-splines quantile objective function on time varying-coefficient modelling. The performances of the proposed techniques is investigated via some simulation studies and applied to crude palm oil (CPO) data as one of the main export commodities in Indonesia. By means of non-crossing p-splines quantile objective function, we model a growth pattern of CPO involving some covariates, such as, rainfall, temperature and humidity.