AMS 588, Failure and Survival Data Analysis
This course introduces both parametric and non-parametric statistical models for analysis
of the failure and survival data – a critical topic in quantitative finance, econometrics,
and biostatistics. Different censoring mechanisms will be discussed. The course will
mainly cover Kaplan-Meier estimator for characterizing the distribution of the failure
and survival data, non-parametric log-rank test for comparing multiple groups, and
the accelerated failure time model and Cox regression model uncovering various predictor/explanatory
variables to survival/failure. Applications to finance, economics and biomedicine
will be illustrated.
We have revised the course title and content to better suit our current graduate programs
in Applied Mathematics and Statistics that have evolved substantially from our old
forms. In our current program, students from many tracks, especially in statistics and in quantitative finance, need this updated course as a highly relevant and important elective. This same
subject is generally referred to as ‘Survival data analysis’ in biostatistics, but
‘Failure data analysis’ in finance. This updated title will reflect the content of
the course clearly for students from all tracks.
3 credits, ABCF grading
Text:
Survival Analysis: Techniques for Censored and Truncated Data, 2nd edition, by John P. Klein and Melvin L. Moeschberger; SAS Inst. Publishing; ISBN: 978-0387953991; and
Survival Analysis Using The SAS System: A Practical Guide, by Paul D. Allison; Springer; ISBN 978-1599946405
THIS COURSE IS OFFERED EVERY TWO OR THREE YEARS.
Learning Outcomes
AMS 588, Failure and Survival Data Analysis
1) Demonstrate skills of working with common problems related to failure data in finance, econometrics, and biostatistics.
- Understand basic settings for problems in failure and survival data;
- Be able to recognize problems related to failure and survival data;
- Analysis of time to event data.
2) Demonstrate skills with statistical inference for failure and survival data.
- Kaplan-Meier estimator for characterizing the distribution of time-to-event data;
- Non-parametric logrank test for comparing multiple groups;
- Accelerated failure time models: exponential, Weibull, log-normal and Gamma;
- Cox regression model.
3) Understand mathematical properties of methods used in survival analysis of failure data.
- Conditional expectation and variance;
- Central limit theorem and delta methods;
- Partial likelihood construction.
4) Demonstrate skills with proficient usage of standard statistical software tools for failure and survival data analysis.
- Understanding of the assumptions, derivation, interpretation of results from survival statistical analysis;
- Proficient in SAS procedures: LIFETEST, LIFEREG and PHREG.