Reinardsyah, Muhammad Irvan (2021) Performa Analisis Regresi Ridge Restricted Dalam Mengatasi Multikolinearitas Pada Data Simulasi. Sarjana thesis, Universitas Brawijaya.
Abstract
Metode Kuadrat Terkecil (MKT) merupakan metode yang paling umum digunakan untuk menduga parameter regresi, namun terdapat beberapa asumsi yang harus dipenuhi agar MKT dapat menghasilkan model yang bersifat BLUE (Best Linear Unbiased Estimator), salah satunya adalah asumsi non-multikolinearitas. Metode yang dapat digunakan untuk mengatasi data yang mengandung multikolinearitas seperti regresi ridge, restricted least square, dan principal component analysis. Pada penelitian ini, digabungkan metode yang mendasari regresi ridge dan restricted least square yang bernama regresi ridge restricted untuk mengatasi
English Abstract
Ordinary Least Square (OLS) is the most commonly used method to estimate regression parameters, but there are several assumptions that must be met so that OLS can produce a BLUE (Best Linear Unbiased Estimator) model, one of which is the assumption of non-multicollinearity. Methods that can be used to address data that contain multicollinearity include ridge regression, restricted least square, and principal component analysis. In this study, the method that underlies the ridge regression and restricted least squares called ridge restricted regression is combined to overcome multicollinearity in the simulation data with sample sizes of 20, 50, and 100, and with correlation coefficients of 0.9, 0.95, and 0.99. The value of the ridge bias constant (c) which is the amount of bias in the estimated value of the regression parameter is determined using the Hoerl, Kennard, & Baldwin (HKB) formula and the Lawless & Wang (LW) formula. Based on the simulation results on various sample sizes and correlation coefficients, the HKB bias constant produces a ridge restricted regression model that is free from multicollinearity with the Mean Square Error (MSE) which is smaller than the LW bias constant. The greater the correlation coefficient in the data, it will increase the MSE produced, while increasing the number of samples will reduce the MSE generated from a model.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | 0521090018 |
| Divisions: | Fakultas Matematika dan Ilmu Pengetahuan Alam > Statistika |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 11 Feb 2022 07:02 |
| Last Modified: | 24 Feb 2022 03:33 |
| URI: | http://repository.ub.ac.id/id/eprint/189687 |
![[thumbnail of DALAM MASA EMBARGO]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text (DALAM MASA EMBARGO)
Muhammad Irvan Reinardsyah.pdf Restricted to Registered users only until 31 December 2023. Download (1MB) |
Actions (login required)
 |
View Item |