In this paper, we introduce different methods for estimating a conditional distribution function, and give two new methods to estimate the conditional distribution function. These two methods all involve empirical-likelihood and forms of N-W estimator. In nonparametric methods, empirical-likelihood is encouraged by its attractive performance, so our estimators based empirical-likelihood have good properties. The first method combines unite moment and the second combines conditional moment, after simulating the second method the performance is good .
|