Attrition rate prediction is a major economic concern for many companies. Different learning approaches have been proposed, however, the a priori choice of the most suitable model remains a non-trivial task as it is highly dependent on the intrinsic characteristics of the churn data. Our study compares eight supervised learning methods combined with seven sampling approaches on thirteen public churn data sets. Our evaluations, reported in terms of area under the curve (AUC), explore the influence of rebalancing and data properties on the performance of learning methods. We rely on the Nemenyi test and Correspondence Analysis as a means of visualization of the associations between models, rebalancing and data. Our comparative study identifies the best methods in an attrition context and proposes a powerful generic pipeline based on an ensemble approach.