My researches of interesting are theoretical development as well as data analysis in the following areas: Sampling plans; Maintenance; Survival Analysis; Quality Control; Bays’ Estimation; Bootstrap Methods; EM algorithm and sampling plans for life-testing in industry. The following are short simply descriptions for each topic.

1. Survival Analysis: Censored data happens very often in medical sciences and industrial life testing. Survival Analysis and related comparisons are important research subjects in medical sciences and industrial life testing. Currently, I am working on the research to develop new method incorporating frailty to analyze and compare life-time data (censoring possibly) from two different treatments; and smooth estimator for Survival Function under random censored data.
2. Quality Control: In industry, many data are not normally distributed. Shewhart type control chart is not suitable under this case. Therefore, new approaches are needed to setup control charts.
3. Bootstrap method is a re-sampling procedure. When the sampling distribution for an estimator (or called estimating function) unknown, bootstrap method usually provides a good approach to estimate sampling distribution for estimator interesting.
4. EM algorithm is an iterative process for finding Maximum Likelihood Estimator (MLE) when the complete likelihood function could not be presented in a closed form or data missing. The most important is to set up an iterated process which contains finding expectation and finding maximization.
5. Maintenance strategy combines preventive repair actions at prescheduled times and minimal repair actions whenever a failure happens. The research contains setting the cost for the strategy and the optimal preventive maintenance schedules to minimize the cost for maintenance strategy and estimating the schedules based on failure times from products. The research could be implemented by simulation for a failure model.
6. Sampling plans is to decide the sampling strategy, which includes sample size and maximal number of nonconforming, such that the producer’s risk and customer’s risk could be met.