Predicting Probability of Customer Churn in Insurance

Abstract

Abstract. We focus on a real case of the motor insurance sector. We propose four different methods to predict lapsing from a portfolio of policies. We present a comparative analysis between three different performance measures in order to assess the predictive power of each model. Our comparison analyses the outcomes of a logistic regression, a conditional tree, a neural network and a support vector machine. These are all considered basic approaches to data mining and knowledge discovery. The main contribution of this paper is to show that, depending on the type of analysis and the objective of the researcher, the optimal prediction method may differ.

Non-life insurance cancellation: a systematic quantitative literature review

“…The publications selected were analysed according to their methodological approach. Eight papers are theoretical (Larsson and Broström 2019;Brophy 2015;Gallo 2014;Van Gelder et al 2018;Guillenet al 2008;Naujoks et al 2017;Rawson et al 2013;Julie Robson 2015) and 18 are empirical; of the latter, 16 are quantitative (Bolancé et al 2016;Brockett et al 2008;de la Llave et al 2019;Felício and Freire 2016;Frank and Lamiraud 2009;Gamble et al 2009;Guillen et al 2003Guillen et al , 2009Guillén et al 2012;Günther et al 2014;Haugen and Moger 2016;Jeong et al 2018;Lin 2010;López-Díaz et al 2017;Paredes 2018;Staudt and Wagner 2018), one is qualitative (Robson and Sekhon 2011) and one uses a mixed methodology (Dominique-Ferreira 2018). This sample indicates that the study of the cancellation of insurance policies is predominantly conducted through empirical quantitative analysis.…”

“…Regarding the methods of data collection for the empirical research, longitudinal analysis predominates, applied in 11 papers (Bolancé et al 2016;Brockett et al 2008;de la Llave et al 2019;Guillen et al 2003Guillen et al , 2009Guillén et al 2012;Günther et al 2014;Haugen and Moger 2016;Jeong et al 2018;Paredes 2018;Staudt and Wagner 2018), over cross-sectional analysis, used in six papers (Frank and Lamiraud 2009;Gamble et al 2009;Lin 2010;Robson and Sekhon 2011;López-Díaz et al 2017), including the paper that used a mixed qualitative and quantitative methodology (Dominique-Ferreira 2018). In the paper by Felício and Freire (2016), both longitudinal and cross-sectional approaches were applied.…”

“…Structural equations were applied in two articles (Felício and Freire 2016;Dominique-Ferreira 2018). Other techniques used in the reviewed documents were probit regression (de la Llave et al 2019), OLS multiple linear regression analyses (Gamble et al 2009), factor analysis (Lin 2010), association rule (Jeong et al 2018) and ROC curves to compare methods to predict lapsing (Bolancé et al 2016).…”

“…Age was analysed to predict product cancellation by Bolancé et al (2016), de la Llave et al (2019), Guillen et al (2003Guillen et al ( , 2008, Paredes (2018) and Staudt and Wagner (2018).…”

“…Insurers' retention strategies have been ineffective, considering their low capacity to strategically retain their best customers (Kofman and Nini 2013). According to Bolancé et al (2016), customer retention is one of the most important priorities for most insurance companies.…”