Upcoming SlideShare
×

# Jurnal Time Series

4,198 views

Published on

1 Like
Statistics
Notes
• Full Name
Comment goes here.

Are you sure you want to Yes No
• Be the first to comment

Views
Total views
4,198
On SlideShare
0
From Embeds
0
Number of Embeds
15
Actions
Shares
0
181
0
Likes
1
Embeds 0
No embeds

No notes for slide

### Jurnal Time Series

1. 1. J. Indones. Math. Soc. (MIHMI) Vol. xx, No. xx (20xx), pp. xx–xx. UJI LINEARITAS TIPE LAGRANGE MULTIPLIER DENGAN EKSPANSI TAYLOR UNTUK DETEKSI HUBUNGAN NONLINEAR PADA DATA TIME SERIES SUBANAR and SUHARTONO Abstract. This paper discusses some latest progress on nonlinear time series analy- sis, particularly about linearity tests that developed based on concepts from theory of neural networks. These statistics tests are for preliminary identiﬁcation whether a non- linear model must be used to analyze a time series. In general, there are two kinds of the neural networks linearity tests which are included a Lagrange Multiplier (LM) test, those are White test and Terasvirta test. Both of these tests are derived from the same single-hidden-layer neural networks. White test is based on the random sampling of the parameter values of neural networks model, whereas Terasvirta test is using Taylor ex- pansion. This research is focused on the Terasvirta test. Here, the theoretical study is considered and also the possibility to develop a new statistics test for linearity using neural networks is discussed. Finally, simulation study is used to evaluate the power of the test and to compare to the result of White test. The result of the simulation study shows that Terasvirta test is more eﬀective than White test to detect nonlinearity in time series. 1. PENDAHULUAN Pada beberapa dekade terakhir ini, pemodelan yang digunakan untuk menje- laskan hubungan nonlinear antar variabel dan beberapa prosedur pengujian untuk Received dd-mm-yyyy, Accepted dd-mm-yyyy. 2000 Mathematics Subject Classiﬁcation: Key words and Phrases: Neural networks, Lagrange Multiplier test, Terasvirta test, nonlinear time series 1