ESTIMASI MODEL REGRESI SEMIPARAMETRIK MENGGUNAKAN ESTIMATOR KERNEL UNIFORM (Studi Kasus: Pasien DBD di RS Puri Raharja)
Authors: Anna Fitriani, I Gusti Ayu Made Srinadi, Made Susilawati
Number of views: 347
Semiparametric regression model approach is a model approach that combines parametric regression models and nonparametric regression. On semiparametric regression, most explanatory variables are parametric and nonparametric others are. Independent variables that satisfy parametric assumptions can be predicted by linear regression analysis method, whereas that does not meet the parametric assumptions alleged by the method nonparametrik.Teknik smoothing (smoothing) nonparametric regression curve on the components used in this study using uniform kernel function. Estimation of optimal semiparametric regression curve is determined by the size of the weight or bandwidth (h) is optimal. Selection of the optimal bandwidth will produce a smooth regression curve estimation in accordance with the pattern data. Selection of the optimum bandwidth is determined based on the criteria that the minimum value of GCV. The purpose of this study was to determine the estimated regression function semiparametric dengue cases using kernel estimators uniform. The response of the data used is old data recovery of patients with Dengue Hemorrhagic Fever (DHF). There are six independent variables such as age (years), body temperature (0C), pulse (beats / min), hematocrit (%), platelets (×〖10〗^3/ul), and duration of fever (day). Age, body temperature, pulse, platelets, and duration of fever is a component of parametric and nonparametric hematocrit is a component. Bandwidth (h) the optimal minimum GCV obtained based on the criteria of 0,005. MSE value is generated using multiple linear regression analysis of 0,031. While the semiparametric regression of 0,00437119.
Keywords: Semiparametric Regression, Kernel, Bandwidth, GCV