Pemodelan Mixture Autoregressive (MAR)Dengan Pendekatan Algoritma EM(Studi Kasus Pada Indeks Harga Saham Nikkei 225) - On Mixture Autoregressive Modelling Using EM Algorithms (Applied In Nikkei 225 Stock Exchange Index Stock Exchange Index)

Historini, Diyah Meriana (2010) Pemodelan Mixture Autoregressive (MAR)Dengan Pendekatan Algoritma EM(Studi Kasus Pada Indeks Harga Saham Nikkei 225) - On Mixture Autoregressive Modelling Using EM Algorithms (Applied In Nikkei 225 Stock Exchange Index Stock Exchange Index). Masters thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
1306201013-Master Thesis.pdf - Published Version

Download (495kB)

Abstract

Berbagai metode peramalan yang didasarkan atas asumsi kenormalan residual telah banyak dikembangkan dalam analisis time series linier. Wong dan Li (2000) menyatakan bahwa dalam kondisi riil, banyak ditemui data time series yang non stasioner dalam mean yang cenderung membawa sifat multimodal. Sehingga dikembangkan suatu model time series non linier yang berkaitan dengan sifat multimodal data yang dikenal dengan model Mixture Autoregressive (MAR). Model ini merupakan suatu model yang terdiri dari mixture K komponen Gaussian Autoregressive (AR). Ada beberapa kelebihan dari model MAR, yaitu mampu mengadaptasi sifat data yang fat tails, leptokurtik, platikurtik dan multimodal serta mampu mengakomodir sifat kemiringan data. Pada penelitian ini dilakukan kajian lebih lanjut berkaitan dengan model MAR dan estimasi parameter dengan menggunakan algoritma EM serta aplikasinya pada data saham Nikkei 225. Adapun model yang diperoleh adalah MAR(3; 3, 3, 3). ============================================================ Various forecasting method based on residual normality assumption have developed in linear and nonlinear time series analysis. Wong and Li (2000) said that in real condition, there are a lot of time series data which are not follow the assumption of non-stationer in mean, couple with multimodality, skewness, and leptokurtic. Recently developed nonlinear time series model, called Mixture Autoregressive (MAR) dealing with some characteristics breaking the normality assumption above, is proposed to be studied here. This model consists of K components Autoregressive Gaussian. This research demonstrates the implementation of EM Algorithm in estimating parameters to model Nikkei 225 Stock Exchange Index stock exchange index. The analysis shows that the data follows MAR (3; 3, 3, 3).

Item Type: Thesis (Masters)
Additional Information: RTSt 519.536 His p
Uncontrolled Keywords: fat-tails, leptokurtik, Mixture Autoregressive, EM Algorithm, fat-tails, leptokurtic, Mixture Normal Autoregressive, EM Algorithm
Subjects: Q Science > QA Mathematics > QA9.58 Algorithms
Divisions: Faculty of Mathematics and Science > Statistics > (S2) Master Theses
Depositing User: ansi aflacha
Date Deposited: 07 Jan 2019 07:35
Last Modified: 07 Jan 2019 07:35
URI: http://repository.its.ac.id/id/eprint/60140

Actions (login required)

View Item View Item