Open AccessRegresi Nonparametrik dengan Pendekatan Deret Fourier pada Data Debit Air Sungai Citarum
Author(s) -
Intaniah Ratur Wisisono,
Ade Irma Nurwahidah,
Yudhie Andriyana
Publication year - 2018
Publication title -
mantik
Language(s) - English
Resource type - Journals
eISSN - 2527-3167
pISSN - 2527-3159
DOI - 10.15642/mantik.2018.4.2.75-82
Subject(s) - fourier series , autoregressive integrated moving average , mathematics , statistics , nonparametric statistics , nonparametric regression , box–jenkins , fourier transform , series (stratigraphy) , time series , mathematical analysis , geology , paleontology
River discharge is one of the factors that affect the occurrence of floods. It varies over time and hence we need to predict the flood risk. Since the plot of the data changes periodically showing a sines and cosines pattern, a nonparametric technique using Fourier series approach may be interesting to be applied. Fourier series can be estimated using OLS (Ordinary Least Square). In a Fourier series, nonparametric regression the level of subtlety of its function is determined by their bandwidth (K). Optimal bandwidth determined using the GCV (Generalized Cross Validation) method. From the calculation results, we have optimal bandwidth which is equal to 16 with R2 is 0.7295 which means that 72.95% of the total variance in the river discharge variable can be explained by the Fourier series nonparametric regression model. Comparing to a classical time series technique, ARIMA Box Jenkins, we obtained ARIMA (1,0,0) with RMSE 83.10 while using Fourier series approach generate a smaller RMSE 50.51.