Abstract
Predictions with distance-based linear and generalized linear models rely upon latent variables derived from the distance function. This key feature has the drawback of adding a non-linearity layer between observed predictors and response which shields one from the other and, in particular, prevents us from interpreting linear predictor coefficients as influence measures. In actuarial applications such as credit scoring or a priori rate-making we cannot forgo this capability, crucial to assess the relative leverage of risk factors. Towards the goal of recovering this functionality we define and study influence coefficients, measuring the relative importance of observed predictors. Unavoidably, due to inherent model non-linearities, these quantities will be local -valid in a neighborhood of a given point in predictor space.
Similar content being viewed by others
References
Boj E, Claramunt MM, Fortiana J (2007) Selection of predictors in distance-based regression. Commun Stat A Theory Methods 36:87–98
Boj E, Delicado P, Fortiana J (2010) Local linear functional regression based on weighted distance-based regression. Comput Stat Data Anal 54:429–437
Boj E, Delicado P, Fortiana J, Esteve A, Caballé A (2012) Local distance-based generalized linear models using the dbstats package for r. Documentos de Trabajo de la Xarxa de Referència en Economia Aplicada (XREAP) 11, http://www.pcb.ub.edu/xreap/aplicacio/fitxers/XREAP2012-11.pdf
Boj E, Caballé A, Delicado P, Fortiana J (2013) dbstats: distance-based statistics (dbstats). http://CRAN.R-project.org/package=dbstats, r package version 1.0.3
Brockman MJ, Wright TS (1992) Statistical motor rating: making effective use of your data. J Inst Actuar 119(3):457–543
Cuadras CM (1989) Distance analysis in discrimination and classification using both continuous and categorical variables. In: Dodge Y (ed) Statistical data analysis and inference. North-Holland, Amsterdam, pp 459–473
Cuadras C, Arenas C (1990) A distance-based regression model for prediction with mixed data. Commun Stat A Theory Methods 19:2261–2279
Cuadras CM, Arenas C, Fortiana J (1996) Some computational aspects of a distance-based model for prediction. Commun Stat B Simul Comput 25:593–609
Davidson A, Hinkley D (1997) Bootstrap methods and their application. Cambridge University Press, New York
Efron B, Tibshirani J (1998) An introduction to the Bootstrap. Chapman and Halls, New York
Esteve A, Boj E, Fortiana J (2009) Interaction terms in distance-based regression. Commun Stat A Theory Methods 38:3498–3509
Flachaire E (1999) A better way to bootstrap pairs. Econ Lett 64:257–262
Flachaire E (2005) Bootstrapping heteroskedastic regression models: wild bootstrap vs. pairs bootstrap. Comput Stat Data Anal 49:361–376
Gower JC (1971) A general coefficient of similarity and some of its properties. Biometrics 27:857–874
Gower J, Harding S (1988) Nonlinear biplots. Biometrika 75:445–455
Haberman S, Rensahw AE (1996) Generalized linear models and actuarial science. J R Stat Soc Ser D (The Statistician) 45(4):407–436
Hallin M, Ingenbleek JF (1983) The Swedish automobile portfolio in 1977. A statistical study. Skand Aktuarietidskr (Scand Actuar J) 83:49–64
MacKinnon J (2002) Bootstrap inference in econometrics. Can J Econ 35(4):615–645
MacKinnon J (2006) Bootstrap methods in econometrics. Econ Rec 82:s2–s18
MacKinnon J (2007) Bootstrap hypothesis testing. Queen’s Economics Department Working Paper 1127
McCullagh P, Nelder JA (1989) Generalized linear models, 2nd edn. Chapman and Hall, London
R Development Core Team (2013) R: a language and environment for statistical computing. Vienna, Austria, http://www.R-project.org/
Author information
Authors and Affiliations
Corresponding author
Additional information
Work supported in part by the Spanish Ministerio de Educación y Ciencia and FEDER, grant MTM2010-17323, and by Generalitat de Catalunya, AGAUR grant 2014SGR152.
Rights and permissions
About this article
Cite this article
Boj, E., Costa, T., Fortiana, J. et al. Assessing the Importance of Risk Factors in Distance-Based Generalized Linear Models. Methodol Comput Appl Probab 17, 951–962 (2015). https://doi.org/10.1007/s11009-014-9415-6
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11009-014-9415-6