., Preatin (2018) Integrasi Model Skedul Migrasi dan Model Poisson dengan Hierarchical Bayes Model [Data Migrasi Hasil SP2010 Propinsi Jawa Timur]. Doctoral thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
06211060017002-Disertation.pdf - Accepted Version Download (9MB) | Preview |
Abstract
Migrasi adalah proses perpindahan individu dari satu wilayah ke wilayah lainnya. Teori migrasi banyak berkembang dari berbagai keilmuan, namun kompleksitas fenomena migrasi tidak dapat dijelaskan dari sudut pandang satu bidang keilmuan. Pendekatan model migrasi lebih menguntungkan karena mampu menggabungkan beberapa variabel berdasarkan beberpa teori migrasi. Dua aspek yang mengikuti proses migrasi, adalah individu dan wilayah, memerlukan pendekatan yang berbeda untuk pemodelannya. Data individu atau data level mikro, memerlukan pemodelan yang dikhususkan untuk melihat karakteristik individu terkait keputusan melakukan migrasi. Sedangkan data wilayah atau data level makro, memerlukan pemodelan yang berbeda pula untuk melihat karakteristik wilayah, baik sebagai daerah asal maupun sebagai daerah tujuan migrasi. Pemodelan migrasi lebih komprehensif jika dilakukan integrasi model level mikro dan level makro.
Model hirarki mampu mengatasi permasalahan penggabungan data level mikro dan makro. Model hirarki dengan metode estimasi Bayesian melalui Markov Chain Monte Carlo (MCMC) dan Gibbs Sampling akan mempermudah penyelesaian model hirarki. Penelitian ini bertujuan melakukan integrasi model hirarki pada level mikro, yaitu individu, dan model level makro, yaitu kabupaten/kota. Model mikro menggunakan model skedul migrasi dan model level makro menggunakan model Poisson Gamma dengan variabel independen berdasarkan penelitian-penelitian sebelumnya. Metode estimasi parameter menggunakan pendekatan full-conditional untuk mengatasi integral komplek.
Aplikasi model dilakukan pada data migrasi Provinsi Jawa Timur hasil Sensus Penduduk 2010 dengan 38 kabupaten/kota. Model skedul migrasi pada model mikro dan model integrasi memberikan hasil yang tidak jauh berbeda berdasarkan nilai R-square (0,88 sampai 0,99) dan Persentase MSE terhadap varians data (0,01% sampai 6%). Model Poisson Gamma menghasilkan model yang lebih baik pada model integrasi dibandingkan model makro berdasarkan Persentase MSE terhadap varians data (0,00002% sampai 0,01% untuk model integrasi dan 4% sampai 46% untuk model makro) . Sehingga model integrasi memberikan analisis komprehensif di level mikro dengan model skedul dan makro dengan model Poisson Gamma dengan hasil yang lebih baik dibandingkan model mikro dan model makro.
===========================================================
Migration is the process of moving individuals from one region to another. The theory of migration much evolved from different sciences, but the complexity of the migration phenomenon cannot be explained from a scientific point of view. The migration model approach is more advantageous because it can combine several variables based on some migration theory. Two aspects that follow the migration process are individuals and regions, requiring different approaches to modeling. The individual data or micro-level data, modeling is required to look at individual characteristics with regard to migration decisions. In macro-level data or region data, different modeling is required to see the characteristics of the region both as a region of origin and as a destination for migration. Migration modeling is more comprehensive if integration is done on micro level and macro level models.
Hierarchical model is able to overcome the problem of merging micro and macro level data. Hierarchical models with Bayesian estimation methods through Markov Chain Monte Carlo (MCMC) and Gibbs Sampling will facilitate the completion of hierarchical models. This study aims to integrate hierarchical model at the micro level ie individual and macro level model that is district/city. The micro-level model uses migration schedule model according to single age migration data and macro-level model using Poisson Gamma model with independent variables based on previous studies. The parameter estimation method uses a full-conditional approach to overcome complex integrals.
Application of the model is done on East Java Province migration data from Population Census 2010 with 38 districts/cities. The migration schedule model on the micro model and integration model gives results that are not much different based on the R-square value (0.88 to 0.99) and the percentage of MSE to the data variance (0.01% to 6%). The Gamma Poisson Model produces a better model on the integration model than the macro model based on the percentage of MSE to the data variance (0.00002% to 0.01% for the integration model and 4% to 46% for the macro model). So the integration model provides a comprehensive analysis at the micro level with a schedule and macro model with a Poisson Gamma model with better results than the micro model and macro model.
| Item Type: | Thesis (Doctoral) |
|---|---|
| Additional Information: | RDSt 519.542 Pre i |
| Uncontrolled Keywords: | Model Skedul Migrasi, Model Poisson Gamma, Hierarchical Bayes Model. |
| Subjects: | H Social Sciences > HA Statistics > HA31.7 Estimation Q Science > QA Mathematics > QA278.2 Regression Analysis. Logistic regression Q Science > QA Mathematics > QA279.5 Bayesian statistical decision theory. |
| Divisions: | Faculty of Mathematics, Computation, and Data Science > Statistics > 49001-(S3) PhD Thesis |
| Depositing User: | Preatin - |
| Date Deposited: | 07 Oct 2020 06:31 |
| Last Modified: | 07 Oct 2020 06:31 |
| URI: | http://repository.its.ac.id/id/eprint/59538 |
Actions (login required)
 |
View Item |