The University of Southampton
University of Southampton Institutional Repository

Time series for early churn detection: using similarity based classification for dynamic networks

Time series for early churn detection: using similarity based classification for dynamic networks
Time series for early churn detection: using similarity based classification for dynamic networks

The success of retention campaigns in fast-moving and saturated markets, such as the telecommunication industry, often depends on accurately predicting potential churners. Being able to identify certain behavioral patterns that lead to churn is important, because it allows the organization to make arrangements for retention in a timely manner. Moreover, previous research has shown that the decision to leave one operator for another, is often influenced by the customer's social circle. Therefore, features that represent the churn status of their connections are usually good predictors of churn when it is treated as a binary classification problem, which is the traditional approach. We propose a novel method to extract time series data from call networks to represent dynamic customer behavior. More precisely, we use call detail records of the customers of a telecommunication provider to build call networks on a weekly basis over the period of six months. From each network, we extract features based on each customer's connections within the network, resulting in individual time series of link-based measures. The time series are then classified using the recently proposed similarity forests method, which we further extend in various ways to accommodate multivariate time series. We show that predicting churn with customer behavior represented by time series is a suitable option. According to our results, the similarity forests method together with some of our proposed extensions, perform better than the one-nearest neighbor benchmark for time series classification. Using a time series of a single feature only, the similarity forests method performs as good as traditional churn prediction methods using more features. In fact, compared to traditional methods, similarity forests based approaches perform better when predicting further in the future, and are therefore better at detecting churn early, allowing organizations to make arrangements for retention in a timely manner.

Call detail records, Churn prediction, Dynamic networks, Multivariate time series, Social networks, Time series classification
0957-4174
55-65
Óskarsdóttir, María
1622b6dd-5d25-4228-9418-a1729e9577e0
Van Calster, Tine
a3f3c605-89ad-4938-856e-c1fdb939a1b6
Baesens, Bart
f7c6496b-aa7f-4026-8616-ca61d9e216f0
Lemahieu, Wilfried
be4bae3f-12b9-417a-91a1-c3c264ffe068
Vanthienen, Jan
6f3d818f-0fce-46fa-966b-160e645caf6d
Óskarsdóttir, María
1622b6dd-5d25-4228-9418-a1729e9577e0
Van Calster, Tine
a3f3c605-89ad-4938-856e-c1fdb939a1b6
Baesens, Bart
f7c6496b-aa7f-4026-8616-ca61d9e216f0
Lemahieu, Wilfried
be4bae3f-12b9-417a-91a1-c3c264ffe068
Vanthienen, Jan
6f3d818f-0fce-46fa-966b-160e645caf6d

Óskarsdóttir, María, Van Calster, Tine, Baesens, Bart, Lemahieu, Wilfried and Vanthienen, Jan (2018) Time series for early churn detection: using similarity based classification for dynamic networks. Expert Systems with Applications, 106, 55-65. (doi:10.1016/j.eswa.2018.04.003).

Record type: Article

Abstract

The success of retention campaigns in fast-moving and saturated markets, such as the telecommunication industry, often depends on accurately predicting potential churners. Being able to identify certain behavioral patterns that lead to churn is important, because it allows the organization to make arrangements for retention in a timely manner. Moreover, previous research has shown that the decision to leave one operator for another, is often influenced by the customer's social circle. Therefore, features that represent the churn status of their connections are usually good predictors of churn when it is treated as a binary classification problem, which is the traditional approach. We propose a novel method to extract time series data from call networks to represent dynamic customer behavior. More precisely, we use call detail records of the customers of a telecommunication provider to build call networks on a weekly basis over the period of six months. From each network, we extract features based on each customer's connections within the network, resulting in individual time series of link-based measures. The time series are then classified using the recently proposed similarity forests method, which we further extend in various ways to accommodate multivariate time series. We show that predicting churn with customer behavior represented by time series is a suitable option. According to our results, the similarity forests method together with some of our proposed extensions, perform better than the one-nearest neighbor benchmark for time series classification. Using a time series of a single feature only, the similarity forests method performs as good as traditional churn prediction methods using more features. In fact, compared to traditional methods, similarity forests based approaches perform better when predicting further in the future, and are therefore better at detecting churn early, allowing organizations to make arrangements for retention in a timely manner.

Text
Time series for early churn detection using similarity based classification for dynamic networks - Accepted Manuscript
Download (799kB)

More information

Accepted/In Press date: 3 April 2018
e-pub ahead of print date: 6 April 2018
Published date: 15 September 2018
Keywords: Call detail records, Churn prediction, Dynamic networks, Multivariate time series, Social networks, Time series classification

Identifiers

Local EPrints ID: 422968
URI: http://eprints.soton.ac.uk/id/eprint/422968
ISSN: 0957-4174
PURE UUID: d862eb02-15c8-4508-849f-32cc43a5a2af
ORCID for Bart Baesens: ORCID iD orcid.org/0000-0002-5831-5668

Catalogue record

Date deposited: 08 Aug 2018 16:31
Last modified: 18 Mar 2024 05:17

Export record

Altmetrics

Contributors

Author: María Óskarsdóttir
Author: Tine Van Calster
Author: Bart Baesens ORCID iD
Author: Wilfried Lemahieu
Author: Jan Vanthienen

Download statistics

Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.

View more statistics

Atom RSS 1.0 RSS 2.0

Contact ePrints Soton: eprints@soton.ac.uk

ePrints Soton supports OAI 2.0 with a base URL of http://eprints.soton.ac.uk/cgi/oai2

This repository has been built using EPrints software, developed at the University of Southampton, but available to everyone to use.

We use cookies to ensure that we give you the best experience on our website. If you continue without changing your settings, we will assume that you are happy to receive cookies on the University of Southampton website.

×