Abstract:Uplift models provide a solution to the problem of isolating the marketing effect of a campaign. For customer churn reduction, uplift models are used to identify the customers who are likely to respond positively to a retention activity only if targeted, and to avoid wasting resources on customers that are very likely to switch to another company. We introduce a Qini-based uplift regression model to analyze a large insurance company's retention marketing campaign. Our approach is based on logistic regression m… Show more
“…Also, in practice, the R-learner is fitted in two stages using cross-fitting to emulate the oracle, thus requiring more observations than S-learners. The methodology introduced in Belbahri et al [2019] attacks parameters estimation and addresses the loss-metric mismatch in uplift regression. It can be seen as an S-learner which estimates the model's parameters in two stages.…”
Section: Related Workmentioning
confidence: 99%
“…Our goal is to regularize the conditional means in order to get a better prediction of the quantity of interest, the uplift. Inspired by the work of , Künzel et al [2019], Belbahri et al [2019], which adapt the optimization problem to the uplift context, we propose to define a composite loss function, which can be separated into two pieces:…”
Section: An Uplift Loss Functionmentioning
confidence: 99%
“…This can be measured by the similarity between the theoretical uplift percentiles of predictions compared with empirical percentiles observed in the data. Maximizing the adjusted Qini coefficient, given in Definition 6, maximizes the Qini coefficient and simultaneously promotes grouping the individuals in decreasing uplift bins, which in turn result in concave Qini curves [Belbahri et al, 2019].…”
Section: Model Evaluationmentioning
confidence: 99%
“…For fair comparison, we use the twin neural architecture. The second method is a two-stage method which is based on a derivative-free optimization of the qadj and imposes sparsity [Belbahri et al, 2019]. We denote this model by Qini-based and our proposed method by Twin µ .…”
Section: Comparison With Benchmark Modelsmentioning
confidence: 99%
“…(c) a Qini-based uplift regression model that uses several LHS structures to search for the optimal parameters (see Belbahri et al [2019] for more details). We denote this model by Qini-based.…”
Uplift is a particular case of conditional treatment effect modeling. Such models deal with cause-and-effect inference for a specific factor, such as a marketing intervention or a medical treatment. In practice, these models are built on individual data from randomized clinical trials where the goal is to partition the participants into heterogeneous groups depending on the uplift. Most existing approaches are adaptations of random forests for the uplift case. Several split criteria have been proposed in the literature, all relying on maximizing heterogeneity. However, in practice, these approaches are prone to overfitting. In this work, we bring a new vision to uplift modeling. We propose a new loss function defined by leveraging a connection with the Bayesian interpretation of the relative risk. Our solution is developed for a specific twin neural network architecture allowing to jointly optimize the marginal probabilities of success for treated and control individuals. We show that this model is a generalization of the uplift logistic interaction model. We modify the stochastic gradient descent algorithm to allow for structured sparse solutions. This helps training our uplift models to a great extent. We show our proposed method is competitive with the state-of-the-art in simulation setting and on real data from large scale randomized experiments.
“…Also, in practice, the R-learner is fitted in two stages using cross-fitting to emulate the oracle, thus requiring more observations than S-learners. The methodology introduced in Belbahri et al [2019] attacks parameters estimation and addresses the loss-metric mismatch in uplift regression. It can be seen as an S-learner which estimates the model's parameters in two stages.…”
Section: Related Workmentioning
confidence: 99%
“…Our goal is to regularize the conditional means in order to get a better prediction of the quantity of interest, the uplift. Inspired by the work of , Künzel et al [2019], Belbahri et al [2019], which adapt the optimization problem to the uplift context, we propose to define a composite loss function, which can be separated into two pieces:…”
Section: An Uplift Loss Functionmentioning
confidence: 99%
“…This can be measured by the similarity between the theoretical uplift percentiles of predictions compared with empirical percentiles observed in the data. Maximizing the adjusted Qini coefficient, given in Definition 6, maximizes the Qini coefficient and simultaneously promotes grouping the individuals in decreasing uplift bins, which in turn result in concave Qini curves [Belbahri et al, 2019].…”
Section: Model Evaluationmentioning
confidence: 99%
“…For fair comparison, we use the twin neural architecture. The second method is a two-stage method which is based on a derivative-free optimization of the qadj and imposes sparsity [Belbahri et al, 2019]. We denote this model by Qini-based and our proposed method by Twin µ .…”
Section: Comparison With Benchmark Modelsmentioning
confidence: 99%
“…(c) a Qini-based uplift regression model that uses several LHS structures to search for the optimal parameters (see Belbahri et al [2019] for more details). We denote this model by Qini-based.…”
Uplift is a particular case of conditional treatment effect modeling. Such models deal with cause-and-effect inference for a specific factor, such as a marketing intervention or a medical treatment. In practice, these models are built on individual data from randomized clinical trials where the goal is to partition the participants into heterogeneous groups depending on the uplift. Most existing approaches are adaptations of random forests for the uplift case. Several split criteria have been proposed in the literature, all relying on maximizing heterogeneity. However, in practice, these approaches are prone to overfitting. In this work, we bring a new vision to uplift modeling. We propose a new loss function defined by leveraging a connection with the Bayesian interpretation of the relative risk. Our solution is developed for a specific twin neural network architecture allowing to jointly optimize the marginal probabilities of success for treated and control individuals. We show that this model is a generalization of the uplift logistic interaction model. We modify the stochastic gradient descent algorithm to allow for structured sparse solutions. This helps training our uplift models to a great extent. We show our proposed method is competitive with the state-of-the-art in simulation setting and on real data from large scale randomized experiments.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.