Open AccessPERBANDINGAN TRANSFORMASI BOX-COX DAN REGRESI KUANTIL MEDIAN DALAM MENGATASI HETEROSKEDASTISITAS
Author(s) -
Ni Wayan Yuni Cahyani,
I Gusti Ayu Made Srinadi,
Made Susilawati
Publication year - 2015
Publication title -
e-jurnal matematika
Language(s) - English
Resource type - Journals
ISSN - 2303-1751
DOI - 10.24843/mtk.2015.v04.i01.p081
Subject(s) - heteroscedasticity , homoscedasticity , statistics , quantile regression , ordinary least squares , power transform , mathematics , linear regression , econometrics , regression analysis , quantile , geometry , consistency (knowledge bases)
Ordinary least square (OLS) is a method that can be used to estimate the parameter in linear regression analysis. There are some assumption which should be satisfied on OLS, one of this assumption is homoscedasticity, that is the variance of error is constant. If variance of the error is unequal that so-called heteroscedasticity. The presence heteroscedasticity can cause estimation with OLS becomes inefficient. Therefore, heteroscedasticity shall be overcome. There are some method that can used to overcome heteroscedasticity, two among those are Box-Cox power transformation and median quantile regression. This research compared Box-Cox power transformation and median quantile regression to overcome heteroscedasticity. Applied Box-Cox power transformation on OLS result ????2point are greater, smaller RMSE point and confidencen interval more narrow, therefore can be concluded that applied of Box-Cox power transformation on OLS better of median quantile regression to overcome heteroscedasticity.