Text mining methods applied to insurance company customer calls: a case study. (English) Zbl 1437.91395

Summary: The purpose of this case study is to develop a process for a U.S. personal lines insurance company to improve its customer service, make call center operations more efficient, and reduce costs by analyzing customer calls. Text mining methods such as topic modeling and sentiment analysis are used to study approximately 10,000 nonclaim customer calls from 2016. Results show the most frequent topics of calls and how customer sentiment differs between topics, which will allow the company to adjust its customer service accordingly.


91G05 Actuarial mathematics
62P05 Applications of statistics to actuarial sciences and financial mathematics
Full Text: DOI


[1] Arun, R.; Suresh, V.; Madhavan, C. E. V.; Murthy., M. N., Advances in knowledge discovery and data mining, On finding the natural number of topics with latent dirichlet allocation: Some observations, 391-402 (2010), Menlo Park, CA: American Association for Artificial Intelligence, Menlo Park, CA
[2] Bauer, D.; Gao, J.; Moenig, T.; Ulm, E. R.; Zhu., N., Policyholder exercise behavior in life insurance: The state of affairs, North American Actuarial Journal, 21, 4, 485-501 (2017) · Zbl 1414.91161
[3] Blei, D. M.; Ng, A. Y.; Jordan, M. I., Latent Dirichlet allocation, Journal of Machine Learning Research, 3, 993-1022 (2003) · Zbl 1112.68379
[4] Blei, D. M.; Lafferty., J. D., A correlated topic model of science, Annals of Applied Statistics, 1, 1, 17-35 (2007) · Zbl 1129.62122
[5] Blei, D. M.; Lafferty., J. D.; Srivastava, A.; Sahami, M., Text mining: Classification, clustering, and applications, Topic models (2009), Chapman & Hall/CRC Press
[6] Blei, D. M.; McAuliffe, J. D., Supervised topic models, 1003, 0783 (2010)
[7] Cao, J.; Xia, T.; Li, J.; Zhang., Y., A density-based method for adaptive LDA model selection, Neurocomputing, 72, 1775-81 (2009)
[8] Deveaud, R.; SanJuan, É.; Bellot., P., Accurate and effective latent concept modeling for ad hoc information retrieval, Document Numérique, 17, 1, 61-84 (2014)
[9] Feinerer, I.; Hornik, K.; Meyer., D., Text mining infrastructure in R, Journal of Statistical Software, 25, 5, 1-54 (2008)
[10] Fellows, I., wordcloud: Word Clouds. R Package Version 2.5 (2014)
[11] Feuerriegel, S.; Proellochs, N., 1.3-2 (2018)
[12] Golden, L. L.; Brockett, P. L.; Ai, J.; Kellison., B., Empirical evidence on the use of credit scoring for predicting insurance losses with psycho-social and biochemical explanations, North American Actuarial Journal, 20, 3, 233-51 (2016) · Zbl 1414.91195
[13] Griffiths, T. L.; Steyvers., M., Finding scientific topics, Proceedings of the National Academy of Sciences, 101, suppl. 1, 5228-35 (2004)
[14] Grün, B.; Hornik., K., topicmodels: An R package for fitting topic models, Journal of Statistical Software, 40, 13, 1-30 (2011)
[15] Henry, E., Are investors influenced by how earnings press releases are written?, Journal of Business Communication, 45, 4, 363-407 (2008)
[16] Hu, M.; Liu, B., Mining opinion features in customer reviews, 755-60 (2004)
[17] Loughran, T.; McDonald., B., When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks, Journal of Finance, 66, 1, 35-65 (2011)
[18] Nikita, M., ldatuning: Tuning of the latent Dirichlet allocation models parameters. R Package Version 0.2.0 (2016)
[19] Ramsay, C. M.; Oguledo., V. I., Optimal disability insurance with moral hazards: Absenteeism, presenteeism, and shirking, North American Actuarial Journal, 19, 3, 143-73 (2015) · Zbl 1414.91228
[20] Rinker, T. W., qdap: Quantitative discourse analysis package. R package version 2.3.0 (2017)
[21] Silge, J.; Robinson, D., tidytext: Text mining and analysis using tidy data principles in R, Journal of Open Source Software, 1, 3 (2016)
[22] Stone, P., Proposal for a server-based inquirer (2004)
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.