ESTIMASI MODEL REGRESI SEMIPARAMETRIK MENGGUNAKAN ESTIMATOR KERNEL UNIFORM (Studi Kasus: Pasien DBD di RS Puri Raharja)
ABSTRACT: Semiparametric
regression model estimation is an estimation that combines both parametric and nonparametric
regression model. In semiparametric regression, some of the variables are parametrics
and the others are nonparametrics. Semiparametric regression is used when
relationship pattern between independent and depentdent variables is half known
and half unknown. Regression curve smoothing technique in nonparametric
components in this study was using uniform kernel function. The optimal
semiparametric regression curve estimation was obtained by optimal bandwidth.
By choosing optimal bandwidth, we would obtain a smooth regression curve
estimation in respect to data pattern. In choosing optimal bandwidth, we use
minimum GCV as a criteria.The purpose of this study was to estimate the
semiparametric regression function of dengue fever case using uniform kernel
estimator. There were 6 independent variables namely age (in years) body temperature
(in Celcius), heartbeat (in times/minutes) hematocryte ratio (in percent),
amount of trombocyte (× 10 3 /ul) and fever duration ( in days). Age, body
temperature, heartbeat, amount of trombosyte and fever duration are parametric
components and hematocryte ration is a nonparametric component. The optimal
bandwidth (h) which was obtained with minimum GCVwas 0,005. The value of MSE
which was obtained by using multiple linear regression analysis was 0,031 and
by using semiparametric regression was 0,00437119.
Penulis: Anna Fitriani, I
Gusti Ayu Made Srinadi, Made Susilawati
Kode Jurnal: jpmatematikadd150283
