ESTIMATING SMOOTHING PARAMETERS USING SPLINE AND KERNEL METHODS IN NONPARAMETRIC REGRESSION (SIMULATION STUDY)
Received: 19.06.2021; Revised: 28.07.2021, Accepted: 21.09.2021, Published Online: 26.10.2021
Hazhar Talaat Abubaker Blbas
Statistics Department, Salahaddin University, Erbil, Iraq, PhD Student in Statistics Department at Salahaddin University, hazhar.abubaker@su.edu.krd
Wasfi Taher Kahwachi
(PhD Supervisor), Tishk International University, Erbil, Iraq, wasfi.kahwachi@tiu.edu.iq
Abstract
The goal of this study is to compare smoothing spline and kernel methods in nonparametric regression models. Thirty-five patients for AML type Leukemia cancer were gathered at Nanakali Hospital for Blood in Erbil, Iraq, from 2015 to 2020. The Bone Marrow (BM) percentage outcome of Leukemia cancer as an explanatory variable and Platelet result as a response variable were used to find out the difference between models.
The results for both real and simulated data indicate that smoothing spline nonparametric regression estimators outperform kernel nonparametric regression estimators using the R language program. Furthermore, increasing the sample size enhances both the Spline and Kernel methods in data with outliers.
Keywords: Nonparametric Regression, Kernel Method, Spline Method, Robust, Leukemia Cancer, AML, BM, PLT