PENDEKATAN REGRESI NONPARAMETRIK DENGAN MENGGUNAKAN ESTIMATOR KERNEL PADA DATA KURS RUPIAH TERHADAP DOLAR AMERIKA SERIKAT

DEWA AYU DWI ASTUTI (Udayana University)
I GUSTI AYU MADE SRINADI (Udayana University)
MADE SUSILAWATI (Udayana University)

E-Jurnal Matematika
Vol 7 No 4 (2018)

Nonparametric regression can be applied for some data types one of them is time series data. The technique of this method is called smoothing technique. There are several smoothing techniques however this study used kernel estimator with seven kernel functions in data of rupiah exchange rate to US dollar. The analysis with R shows that by using minimum Generalized Cross Validation (GCV) criteria, seven functions produce various optimal bandwidth value but has similar curves estimation. The conclusion is that by using kernel estimator in time series data support that choosing the optimal bandwidth is more important than choosing the kernel functions.