One way to overcome of outliers in the spatial regression model is by using robust spatial regression. Removing outliers in spatial analysis can change the composition of spatial effects on data. One reason for the inaccuracy of the spatial regression model in predicting is the existence of outlier observations. If there are spatial influences on both variables, the model that will be formed is Spatial Durbin Model. Spatial regression is a model used to determine relationship between response variables and predictor variables that gets spatial influence. The influence value of predictor variabels to poverty is 73,08% and the MSE value is 0,00791. The best ridge robust regression model is Ridge Robust Regression S-Estimator. Applied the ridge robust regression to case study prove that ridge robust regression can improve parameter estimation. The result of estimation using OLS show that there is a multicollinearity and presence an outliers. The case study can be used with this method is data with multicollinearity and outlier, the case study in this research is poverty in Central Java 2020 influenced by life expentancy, unemployment number, GRDP rate, dependency ratio, human development index, the precentage of population over 15 years of age with the highest education in primary school, mean years school. Ridge Robust Regression is a modification of ridge regression method used to solve the multicolliniearity and using some estimators of robust regression used to solve the outlier, the estimator including : Maximum likelihood estimator (M-estimator), Scale estimator (S-estimator), and Method of moment estimator (MM-estimator). One method can be used to solved the multicollinearity and outliers problem is Ridge Robust Regression. If there is a violation of assumptions such as multicolliniearity especially coupled with the outliers, then the regression with OLS is no longer used. The Ordinary Least Squares (OLS) is one of the most commonly used method to estimate linier regression parameters.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |