Estimasi Parameter Pada Model Poisson Generalized Autoregressive Moving Average (GARMA) Dengan Algoritma IRLS (Studi Kasus: Peramalan Jumlah Kecelakaan Di Jalan Tol Surabaya-Gempol) - Parameter Estimation Of Poisson Generalized Autoregressive Moving Average (GARMA) Models With IRLS Algorithm (A Case study is forecasting of Accident in Surabaya-Gempol Toll Road)

Fauzia, Agil Desti (2018) Estimasi Parameter Pada Model Poisson Generalized Autoregressive Moving Average (GARMA) Dengan Algoritma IRLS (Studi Kasus: Peramalan Jumlah Kecelakaan Di Jalan Tol Surabaya-Gempol) - Parameter Estimation Of Poisson Generalized Autoregressive Moving Average (GARMA) Models With IRLS Algorithm (A Case study is forecasting of Accident in Surabaya-Gempol Toll Road). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[img] Text
06111440000062-Draft.pdf - Published Version
Restricted to Repository staff only

Download (1MB) | Request a copy

Abstract

Peramalan adalah pengolahan data masa lalu untuk mendapatkan estimasi data masa depan. Data yang digunakan pada penelitian ini adalah data count. Pada kasus data count metode peramalan pada umumnya seperti ARIMA kurang tepat digunakan. Benjamin, dkk. mengembangkan sebuah model peramalan yaitu Generalized Autoregressive Moving Average (GARMA) dengan menggunakan fungsi penghubung (link function) dengan data diasumsikan mengikuti Distribusi Poisson sehingga disebut juga Poisson GARMA (p,q). Pada model tersebut terdapat beberapa parameter yang tidak diketahui. Parameter yang dimaksud diestimasi menggunakan metode Maximum Likelihood Estimation (MLE) dengan optimasi Algoritma Iteratively Reweighted Least Squares (IRLS). Model Poisson GARMA ini diterapkan pada data jumlah kejadian kecelakaan di jalan tol Surabaya-Gempol ruas Waru-Sidoarjo. Hasil yang didapat yaitu model khusus Poisson GARMA (1,1) dengan 3 parameter yaitu parameter konstanta , parameter Autoregressive, dan parameter Moving Average . Kriteria pemilihan model terbaik menggunakan AIC.====================================================================================================================Forecasting is the processing of past data to obtain future data estimates. Data used in this research is data count. In the case of data count, forecasting methods in general such as ARIMA is less precise for some reasons. Benjamin, et al. developed a forecasting model to solve this, namely Generalized Autoregressive Moving Average (GARMA) by using the link function. The data used is assumed to follow Poisson's distribution so it is also called Poisson GARMA (p, q). In the model, there are some unknown parameters. These parameters are estimated using Maximum Likelihood Estimation (MLE) method with Iteratively Reweighted Least Squares (IRLS) algorithm optimization. Poisson GARMA model is applied to the data of the number of accidents on the Surabaya-Gempol toll road at Waru-Sidoarjo section. The result obtained is a special model Poisson GARMA (1,1) with 3 parameters namely constant parameters, Autoregressive parameter, and Moving Average parameter. The criteria of best model selection uses AIC.

Item Type: Thesis (Undergraduate)
Additional Information: RSMa Fau e
Uncontrolled Keywords: Data count, Distribusi Poisson, Fungsi Link, Poisson GARMA (
Subjects: Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis.
Divisions: Faculty of Mathematics, Computation, and Data Science > Mathematics > (S1) Undergraduate Theses
Depositing User: Agil Desti Fauzia
Date Deposited: 28 Dec 2018 03:40
Last Modified: 28 Dec 2018 03:40
URI: http://repository.its.ac.id/id/eprint/58866

Actions (login required)

View Item View Item