컨텐츠 시작
학술대회/행사
초록검색
제출번호(No.) | 0575 |
---|---|
분류(Section) | Poster Session |
분과(Session) | Probability / Stochastic Process / Statistics (SS-12) |
영문제목 (Title(Eng.)) |
[CANCELLED] A comparison of methods for generating event history data |
저자(Author(s)) |
Danardono1 Department of Mathematics, Universitas Gadjah Mada1 |
초록본문(Abstract) | Event history data or time-to-event data frequently arises in many scientific investigations. To analyze event history data, the Cox proportional hazards model and its extension are usually used. Theoretical and practical developments in the Cox proportional hazards model are usually investigated by means of simulation studies. Therefore, valid and efficient methods to generate data that follow the Cox proportional hazards is indispensable. There are several methods for generating such event history data, three of them are inverse methods, permutational algorithms and methods based on rank. In this paper, the three methods are compared by conducting several simulation scenarios to assess the accuracy, stability, easiness of implementation and computing time. The inverse methods have considerably small computing time but it is only possible for parametric proportional hazards models. The permutational algorithms and the method based on rank are generally flexible to implement although the computing time is slower as the events increased. The permutational algorithm and the method based on rank are recommended for general implementation of generating event history data. |
분류기호 (MSC number(s)) |
62N99 |
키워드(Keyword(s)) | survival analysis, simulations, proportional hazards model, time-to-event data |
강연 형태 (Language of Session (Talk)) |
English |