Abstract
In this paper we propose a novel procedure, for the estimation of semiparametric survival functions. The proposed technique adapts penalized likelihood survival models to the context of lifetime value modeling. The method extends classical Cox model by introducing a smoothing parameter that can be estimated by means of penalized maximum likelihood procedures. Markov Chain Monte Carlo methods are employed to effectively estimate such smoothing parameter, using an algorithm which combines Metropolis–Hastings and Gibbs sampling. Our proposal is contextualized and compared with conventional models, with reference to a marketing application that involves the prediction of customer’s lifetime value estimation.
Similar content being viewed by others
References
Aalen O (1989) A linear regression model for the analysis of life times. Stat Med 8: 907–925
Cox DR (1972) Regression models and life tables. J R Stat Soc Series B 34: 187–220
Figini S (2007) Bayesian variable and model selection for customer lifetime value, PhD dissertation, University Bocconi, Milan
Figini S, Giudici P (2009) Bayesian churn models. To Appear in Adv Appl Stat Sci
Figini S, Giudici P, Galliano D, Polla D (2007) Survival analysis models to estimate customer lifetime value. IEEE Press, New York
Grambsch TM, Therneau PM (2001) Modeling survival data: extending the Cox model. Springer, New York
Hougaard P (1995) Frailty model for survival data. Lifetime Data Anal 1: 255–273
Ibrahim JG, Chen MH, Sinha D (2001) Bayesian survival analysis. Springer, New York
Kalbfleisch JD (1978) Nonparametric Bayesian analysis of survival time data. J R Stat Soc Series B 40: 214–221
Madigan D, Hoeting JA, Raftery AE, Volinsky CT (1999) Bayesian model averaging. A tutorial. Stat Sci 14: 382–417
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Figini, S. Penalized models to estimate customer survival. Stat Methods Appl 19, 141–150 (2010). https://doi.org/10.1007/s10260-009-0126-z
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10260-009-0126-z