Master's degree in Statistics and Operations Research

Share:

Statistical Methods for Lifetime Data

Títol del curs: Statistical Methods for Lifetime Data

Impartit per: Luís A. Escobar. Department of Experimental Statistics, Louisiana State University.

Llengua del curs: Castellà

Dates i horaris del curs: 13-15 de juny 18-19 de juny 2012. 16-20h.

Lloc: PC2

Tipus d'activitat i càrrega lectiva: Curs de 20 hores

Reconeixement acadèmic: 2.5 crèdits

Data de matrícula: del 15 al 25 de maig

Course Description:
This is an applied course covering advanced statistical methods for lifetime data analysis; in particular, as they apply to Reliability and Survival analysis studies. Reliability/Survival improvements require timely decisions based on censored data. The methods described in this course are important tools for analyzing reliability/survival data. The course will focus on concepts, examples, models, data analysis, and interpretation. Assessment will be made by exercises or case studies proposed by the teacher.

After completing this course, you will be able to:

  • Recognize and properly deal with different kinds of reliability/survival data and properly interpret important reliability/survival metrics.
  • Use nonparametric estimation to make inferences from multiply censored data with minimal assumptions.
  • Use probability plots to identify appropriate parametric models and diagnose anomalies in failure-time data.
  • Fit simple semi-parametric models including the standard techniques required when using proportional hazard models.
  • Fit simple models to life data, and make inferences on important quantities like distribution quantiles, failure probabilities, and hazard functions.
  • Use appropriate methods for computing confidence intervals from censored data.
  • Identify and analyze data with multiple failure modes.
  • Use regression analysis methods for the analysis of nonnormal-censored data that arise in field studies and in controlled life studies.
  • Understand the relationship and tradeoffs between traditional failure-time data and degradation data.
  • Conduct comparative studies using lifetime data.

Course Outline:

Part 1 (6 hours):

  • Appropriate Time Scales for Life Data/Degradation Data.
  • Lifetime Metrics: Failure Probability, Quantiles, Hazard.
  • Simple Nonparametric Estimation, the Kaplan Meier Estimator.
  • Introduction to Software (JMP/R).
  • Weibull/Lognormal Distributions.

 

Part 2 (6 hours):

  • Probability Plots, Detecting Multiple Modes of Failure.
  • Parametric Modeling with Single Distribution.
  • Multiple Failure Modes and Data Analysis.
  • Semi-parametric Regression Analysis.
  • Fitting the Proportional Hazard Regression Model.

 

Part 3 (5 hours):

  • Failure Time Regression Analysis.
  • Principles of Acceleration Models and Acceleration Factors.
  • Regression Life Test Data Analysis--One Variable.
  • Regression Life Test Data Analysis--More than One Variable.
  • Pitfalls in Accelerated/Regression Test Models.

 

Part 4 (3 hours):

  • Accelerated Repeated Measures Degradation.
  • Comparisons.
  • Accelerated Destructive Degradation Test Data Analysis.

Presentació:
Luis is a Professor in the Department of Experimental Statistics, Louisiana State University. He holds a BS from National University, Medellin, Colombia, a MS from the Inter-American Statistical Training Center (CIENES), Santiago, Chile, and a Ph.D. from Iowa State University.
Luis is President Elect of the Inter-American Statistical Institute (IASI). His research and consulting interests include statistical analysis of reliability data, accelerated testing, survival analysis, linear and non-linear models. Professor Escobar is past Associate Editor for Lifetime Data Analysis and past Associate Editor for Technometrics. He is a Fellow of the American Statistical Association and an elected member of the International Statistical Institute. Professor Escobar was awarded the 1999 Jack Youden Prize and he has won two awards for outstanding teaching at Louisiana State University. In 2009 Luis was awarded the Rainmaker distinction at Louisiana State University because he was considered one of the top 100 researchers in the University among a faculty of 3000 members. Luis is the co-author of Statistical Methods for Reliability Data (Wiley 1998), and several other book chapters. His publications have appeared in engineering and statistical literature.