Ensembles of Probability Estimation Trees for Customer Churn Prediction

Abstract

Customer churn prediction is one of the most important elements of any Customer Relationship Management (CRM) strategy. In this study, a number of strategies are investigated to increase the lift of ensemble classification models. In order to increase lift performance, two elements of a number of well-known ensemble strategies are altered: (i) the potential of using probability estimation trees (PETs) instead of standard decision trees as base classifiers is investigated and (ii) a number of alternative fusion rules for the combination of ensemble member's outputs are proposed and compared. Experiments are conducted for four popular ensemble strategies (Bagging, the Random Subspace Method, SubBag and AdaBoost) on five real-life churn data sets. The results demonstrate the value of using PETs over standard decision trees in order to increase lift. Overall, the effect of the proposed strategies heavily depends on the chosen ensemble algorithm in which they are implemented.

A SLR on Customer Dropout Prediction

“…The dropout research mainly addresses telecommunications, which corresponds to approximately 44% of the publications, followed by the financial area (representing 18%); the remaining research focuses on the areas of software (3%), retail (2%) and games (2%), while energy, gambling, hospitality, logistics and security represent only 1% each (Figure 4). There are overlapping areas that target several business areas, such as the financial and telecom industries (6%) [9,29,46,90,122], financial and retail industries(3%) [68,74,97], financial, telecom and retail industries (1%) [42], and telecom and media industries (1%) [18]. These studies developed an analysis targeting more than one business area as proof of concept of the approach being tested to predict dropout.…”

“…Another ensemble method was bagging (bootstrap + aggregating), which adopts a model averaging approach and boosting using weak learners to iteratively learn and create a strong classifier. The analyzed studies explored the use of boosting using other approaches, such as (1) AdaBoost [77,97,99,106,116,121]; (2) gradient boosting [28]; and…”

“…Al-Molhem et al [68] conducted social network analysis to enhance the results of churn prediction models in the telecom domain using call detail records to construct a weighed graph representing the distance between two subscribers to calculate the centrality. The usage of marketing related variables, such as promotions offered to a customer, calls developed in a retention strategy, and helpdesk interactions, were applied by Verbeke et al [50] However, certain studies did not identify the features employed [9,12,18,22,25,42,48,67,70,73,85,87,88,97,98,99,107,110,112,113,116,119,122,127], which in some cases are related to the usage of more than one database. Idris et al [119] used two databases (orange and cell2cell).…”

“…Burez and Van den Poel [88] used data from two banks with 117808 and 102279 customers, a mobile dataset (n=100205), newspaper (n=122189) and PayTV (143198). De Bock and Van den Poel [97] explored two datasets (bank1=23562 and bank2=42783), which made them difficult to describe in a research paper.…”

“…The problem of determining customer dropout was also addressed using a perspective to target retention actions [93]; the lift metric allows us to focus on customers with a level of certainty that dropout will occur; and marketing efforts are directed to those who will almost definitively churn. Several studies adopted the lift [5,15,18,43,49,54,78,80,82,89,97,118,125] and accuracy considering that only 10% of customers are most likely to churn [8] in the studies [24,28,29,42,46,90,91,93,95,123,127]. This approach was expanded, and the maximum profit was considered to optimize the targeting action in retention strategies [110].…”