Generalized entropy optimization methods for survival data

Creative Commons License

Shamilov A., Özdemir S., Yılmaz N.

ALT2014 : 5th International Conference on Accelerated Life Testing and Degradation Models, 11 - 13 Haziran 2014 Pau (France), Pau, France, 11 - 13 June 2014, pp.174-183

  • Publication Type: Conference Paper / Full Text
  • City: Pau
  • Country: France
  • Page Numbers: pp.174-183
  • Van Yüzüncü Yıl University Affiliated: Yes


In this research we have developed MinMaxEnt, MaxMaxEnt and MinMinxEnt, MaxMinxEnt methods representing statistical distributions of the same name generated by Shannon entropy measure and KullbackLeibler measure respectively and the finite set of characterizing moment functions. Further, we illustrate the use of these methods on survival data analysis. The performances of improved methods are established by Chi-Square criteria, Root Mean Square Error (RMSE) criteria, Shannon entropy measure and Kullback-Leibler measure. According to obtained distributions (𝑀𝑎𝑥𝑀𝑎𝑥𝐸𝑛𝑡)4 , (𝑀𝑖𝑛𝑀𝑎𝑥𝐸𝑛𝑡)4 , (𝑀𝑖𝑛𝑀𝑖𝑛𝑥𝐸𝑛𝑡)4 and (𝑀𝑎𝑥𝑀𝑖𝑛𝑥𝐸𝑛𝑡)4 estimator of Probability Density Function 𝑓̂(𝑡), Cumulative Distribution Function 𝐹̂(𝑡), Survival Function 𝑆̂(𝑡) and Hazard Function ℎ̂(𝑡) are evaluated and graphically illustrated. The results are acquired by using statistical software MATLAB.