Kernel Estimation of the Conditional Density Function with Application of the Palestinian Merchandise Imports Data

Abstract

The relationship between a current observation and previous observations, where the conditional density function plays an important role, is the main subject of this paper. To study this relation, the unknown conditional density function must be estimated.  In this paper, the kernel estimation of the mean and mode of the conditional density function will be studied, and the conditions under which these estimators are asymptotically normally distributed will be discussed.

The performance of the kernel estimator of the conditional mean and mode will be tested using simulated data. We will analyze the monthly data of the Palestinian merchandise imports using the kernel techniques to predict future observations.

Also, the Box and Jenknis methodology will be applied to the data and its results will be compared to that of the kernel techniques.