Some NONPARAMETRIC ESTIMATORS FOR RIGHT CENSORED SURVIVAL DATA

Abstract

The using of the parametric models and the subsequent estimation methods require the presence of many of the primary conditions to be met by those models to represent the population under study adequately, these prompting researchers to search for more flexible parametric models and these models were nonparametric, many researchers, are interested in the study of the function of permanence and its estimation methods, one of these non-parametric methods.

For work of purpose statistical inference parameters around the statistical distribution for life times which censored data , on the experimental section of this thesis has been the comparison of non-parametric methods of permanence function, the existence of surveillance (Type Ι- censored data) employing simulation style using (Kaplan-Meier estimation method, Kernel estimation method, Nelson-Aalen estimation method, Thompson–Type estimation method and Pandey estimation method) these methods have the most flexibility in data analysis the statements with no knowledge of the distribution who inserts data for the estimator, to get the best way to assess the permanence function using the simulation method of two of the statisticians measure, (IMSE) Integral Mean Squares Error and Mean Absolute Percentages Error (MAPE) for different sample size like (n  =  15,  30,   50  ,100   ), has been reached to a preference of the best way to estimate is Kaplan-Meier method from the remainder of the nonparametric methods, the results show  that  permanence function values  start to  decrease with  increasing  of time  in  relation  to  nonparametric estimation.