Fadliana, Alfi (2019) Pemodelan Locally Compensated Ridge Geographically Weighted Regression (Studi Kasus: Stunting Di Provinsi Bali dan Kepulauan Nusa Tenggara). Magister thesis, Universitas Brawijaya.
Abstract
Stunting adalah kondisi gagal tumbuh pada anak usia di bawah lima tahun akibat kekurangan gizi kronis dan infeksi berulang terutama dalam 1000 HPK (Hari Pertama dalam Kehidupan), yang terlihat dari panjang atau tinggi badan berada di bawah rata-rata panjang atau tinggi anak seusianya. Prevalensi stunting sangat tinggi di hampir semua provinsi di Indonesia sehingga perlu kiranya untuk dilakukan penanganan yang serius. Pendekatan model Geographically Weighted Regression (GWR) untuk mengkaji faktor-faktor yang mempengaruhi prevalensi stunting bisa jadi merupakan pilihan yang tepat karena mampu mengatasi keragaman/heterogenitas spasial. Namun, terkait dengan beberapa faktor yang diduga berpengaruh terhadap prevalensi stunting seperti di antaranya lokasi tempat tinggal (geografis), kondisi ibu, kondisi bayi/Balita, kondisi lingkungan rumah tangga, perilaku hidup bersih, kualitas Sumber Daya Manusia (SDM), dan tingkat perekonomian yang sangat memungkinkan saling berkorelasi atau berhubungan linear di setiap wilayah, penggunaan model GWR akan menjadi kurang efektif. Sebab, GWR mengabaikan dependensi yang berpotensi terjadi pada koefisien regresi lokal antara variabel prediktor yang berbeda, atau yang disebut dengan multikolinearitas lokal. Pada regresi spasial, multikolinearitas lokal dapat diatasi dengan menggunakan konsep dari metode regresi ridge ke dalam GWR yang kemudian dikenal dengan istilah Geographically Weighted Ridge Regression (GWRR). Kelemahan dari model GWRR adalah penambahan satu koefisien bias global,
English Abstract
Stunting is a condition of failure to thrive in children under five years of age due to chronic malnutrition and recurrent infections, especially in the First 1000 Days in Life, which can be seen from the length or height below the average length or height of a child of his age. The prevalence of stunting is very high in almost all provinces in Indonesia so it is necessary to take serious treatment. The Geographically Weighted Regression (GWR) model approach to assessing the factors that influence the prevalence of stunting may be the right choice because it can overcome spatial heterogeneity. However, it is related to several factors that are thought to influence the stunting prevalence such as the location of residence (geographical), the condition of the mother, the condition of infants/toddlers, household environmental conditions, clean living behavior, the quality of human resources, and the level of the economy which is very possible to correlate with each other or be linearly related in each region, the use of the GWR model will be less effective. Because, GWR ignores potential dependencies in the local regression coefficient between different predictor variables, or what is called local multicollinearity. In spatial regression, local multicollinearity can be overcome by using the concept of the ridge regression method into GWR which became known as Geographically Weighted Ridge Regression (GWRR). The disadvantage of the GWRR model is the addition of one global bias coefficient,
Other obstract
-
| Item Type: | Thesis (Magister) |
|---|---|
| Identification Number: | TES/519.536/FAD/p/2019/041906300 |
| Uncontrolled Keywords: | REGRESSION ANALYSIS, REGRESSION ANALYSIS--MATHEMATICAL MODELS |
| Subjects: | 500 Natural sciences and mathematics > 519 Probabilities and applied mathematics > 519.5 Statistical mathematics > 519.53 Descriptive statistics, multivariate analysis, analysis of variance and covariance > 519.536 Regression analysis |
| Divisions: | S2/S3 > Magister Statistika, Fakultas MIPA |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 19 Dec 2019 06:57 |
| Last Modified: | 21 Oct 2021 06:45 |
| URI: | http://repository.ub.ac.id/id/eprint/177178 |
Preview |
Text
Alfi Fadliana (2).pdf Download (4MB) | Preview |
Actions (login required)
 |
View Item |