Evaluation Age and Gender for General Census of the population in Iraq by using nonparametric Bayesian Kernel Estimators

Abstract

The process of evaluating data (age and the gender structure) is one of the important factors that help any country to draw plans and programs for the future. Discussed the errors in population data for the census of Iraqi population of 1997. targeted correct and revised to serve the purposes of planning. which will be smoothing the population databy using nonparametric regression estimator (Nadaraya-Watson estimator) This estimator depends on bandwidth (h) which can be calculate it by two ways of using Bayesian method, the first when observations distribution is Lognormal Kernel and the second is when observations distribution is Normal Kernel.

then we will be compare between the result of these methods by using UN Age-Sex Accuracy Index and analysis of the Age and Gender ratios to find the method which gave the optimum smoothing for data. And we reached that the method of estimate h when observations distributed as Lognormal Kernel of Bayesian method is the best because it achieved less value of UN Age-Sex Accuracy Index.