Zahara, Chasna (2024) Evaluasi Perbandingan Model GSTAR dan GSTAR Poisson pada Peramalan Jumlah Hari Hujan Tertinggi di Jawa Timur. Other thesis, Institut Teknologi Sepuluh Nopember.
Text
5002201022-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy |
Abstract
Kajian mengenai model matematika dalam bidang peramalan terus berkembang. Model peramalan yang mempertimbangkan keterkaitan antara waktu dan lokasi adalah model Generalized Space Time Autoregressive (GSTAR) dimana model ini memiliki dugaan bahwa parameter autoregressive pada setiap lokasi bervariasi sehingga fleksibel untuk lokasi yang heterogen. Namun, model GSTAR masih mengasumsikan bahwa variabel respon berdistribusi Normal dan stasioner sehingga model GSTAR Poisson diusulkan untuk mengatasi data jumlahan dan berdistribusi Poisson. Penelitian ini bertujuan untuk membandingkan keakuratan dua model, yaitu GSTAR dan GSTAR Poisson dalam meramalkan jumlah hari hujan tertinggi di Jawa Timur. Data yang digunakan pada penelitian ini adalah data spasial jumlah hari hujan di Kota Batu, Kota Malang, dan Kota Pasuruan dari tahun 2014 hingga 2023. Dalam membangun model GSTAR, penelitian ini menggunakan bobot lokasi dengan invers jarak untuk merepresentasikan bentuk keterkaitan antar lokasi. Penelitian ini menunjukkan bahwa model GSTAR memiliki nilai RMSE yang lebih kecil daripada model GSTAR Poisson sehingga dapat disimpulkan bahwa model GSTAR mampu meramalkan jumlah hari hujan tertinggi di Jawa Timur lebih akurat dibandingkan dengan model GSTAR Poisson.
============================================================
The study on mathematical models in the field of forecasting continues to evolve. A forecasting model that considers the interrelation between time and location is the Generalized Space Time Autoregressive (GSTAR) model. This model assumes that the autoregressive parameters at each location vary, making it flexible for heterogeneous locations. However, the GSTAR model still assumes that the response variable is normally distributed and stationary. Therefore, the Poisson GSTAR model is proposed to handle count data that follows a Poisson distribution. This study aims to compare the accuracy of two models, namely GSTAR and Poisson GSTAR, in forecasting the highest number of rainy days in East Java. The data used in this study are spatial data on the number of rainy days in Batu City, Malang City, and Pasuruan City from 2014 to 2023. In constructing the GSTAR model, this study uses location weights with inverse distance to represent the interrelation between locations. The study shows that the GSTAR model has a smaller RMSE value than the Poisson GSTAR model, leading to the conclusion that the GSTAR model can more accurately forecast the highest number of rainy days in East Java compared to the Poisson GSTAR model.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | GSTAR, GSTAR Poisson, Jumlah Hari Hujan, Peramalan. Forecasting, GSTAR, Number of Rainy Days, Poisson GSTAR. |
Subjects: | Q Science > QA Mathematics > QA276 Mathematical statistics. Time-series analysis. Failure time data analysis. Survival analysis (Biometry) Q Science > QA Mathematics > QA401 Mathematical models. |
Divisions: | Faculty of Science and Data Analytics (SCIENTICS) > Mathematics > 44201-(S1) Undergraduate Thesis |
Depositing User: | Chasna Zahara |
Date Deposited: | 02 Aug 2024 07:30 |
Last Modified: | 02 Aug 2024 07:30 |
URI: | http://repository.its.ac.id/id/eprint/112126 |
Actions (login required)
View Item |