Comparison of quantile regression to lognormal and gamma regression using birth weight data

Loading...
Thumbnail Image
Date
November,2015
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Over the years, positively skewed data such as data from insurance, economics, laboratory, health and so on, have been analysed using conditional mean models such as simple linear regression and logistic regression. Estimation of these models can be seriously deficient if constructed on some non-gaussian settings and cannot be readily extended to non-central location which is precisely where the interest of a social science research often reside. This study therefore seeks to employ a methodology to deal with these problems. This study seeks to estimate the quantiles that describe the entire distribution and also to obtain an appropriate statistical distribution for the birth weight data. Our study used birth weight data from Komfo Anokye Teaching Hospital. Quantile, Lognormal and Gamma regression were used in the analysis and Quantile-Quantle plot and Akaike’s Information Criterion(AIC) were the goodness of fit test for the selection of the distribution that fitted the data well. th th th th th Finally we estimated 5 , 25 , 50 , 75 and 95 quantile regression to describe the entire distribution of the data. The lognormal also was selected as a better distribution than the gamma distribution based on their AIC values and the graph of the Q-Q plot.
Description
A thesis submitted to the Department of Mathematics, Kwame Nkrumah University of Science and Technology in partial fulfillment of the requirement for the Degree of Master of Philosophy in Mathematical Statistics,
Keywords
Citation