Institutional Repository UIN Sunan Kalijaga Yogyakarta: No conditions. Results ordered -Date Deposited. 2024-03-28T20:45:33ZEPrintshttp://digilib.uin-suka.ac.id/images/sitelogo.pnghttps://digilib.uin-suka.ac.id/2015-12-18T03:09:10Z2015-12-18T03:09:10Zhttp://digilib.uin-suka.ac.id/id/eprint/18718This item is in the repository with the URL: http://digilib.uin-suka.ac.id/id/eprint/187182015-12-18T03:09:10ZPENERAPAN GLOBAL RIDGE-REGRESSION PADA PERAMALAN DATA TIME SERIES NON LINEAR STUDI KASUS : PEMODELAN NILAI TUKAR US DOLAR TERHADAP RUPIAHTime series modelling has two types i.e. linear and non-linear. Feed Forward
Neural Networks (FFNN) has modelled linear time series well but has found difficulties
to model non-linear time series. Radial Basis Function Neural Networks (RBFNN) give
an alternative to model non-linear time series. This network has Radial Basis Function
in the hidden layer that provides non-linear functions. The RBFNN output is a linear
combination of Radial Basis Functions and output weights. An optimal output has the
least square error. The weights are gotten from regression. Global-ridge regression
adds a regulation parameter to give the optimal weights that produce an optimal
output.
Keyword : Radial Basis Function Neural Networks (RBFNN), non-linear, time series, global ridge-regressionSri Utami ZULIANA