NON-PARAMETRIC STATISTICAL INFERENCE FOR THE SURVIVAL EXPERIMENTS

Generally, in most of the real life experiments, all the individuals are not attaining the specified event of interest within the stipulated time period and they will be contributing to the censored data and even in some studies, the amount of censoring seems to be very heavy. It is observed that the celebrated Kaplan-Meier Estimator (KME) provides overestimates in such heavy censoring situations and so to deal this problem of overestimation, a new weight has been designed and its efficiency was attained empirically through the Improved Weighted Kaplan-Meier Estimator (IWKME). In this paper, the extension of KME was implemented empirically on the study data as well as on the modified datasets and its corresponding conclusion has been drawn. Also, the author has completed this study with its curious conclusions by comparing the analysis of the log-rank test and weighted log-rank tests with the proposed weighted log-rank tests.