A simple plug-in bagging ensemble based on threshold-moving for classifying binary and multiclass imbalanced data

Abstract

Class imbalance presents a major hurdle in the application of classification methods. A commonly taken approach is to learn ensembles of classifiers using rebalanced data. Examples include bootstrap averaging (bagging) combined with either undersampling or oversampling of the minority class examples. However, rebalancing methods entail asymmetric changes to the examples of different classes, which in turn can introduce their own biases. Furthermore, these methods often require specifying the performance measure of interest a priori, i.e., before learning. An alternative is to employ the threshold moving technique, which applies a threshold to the continuous output of a model, offering the possibility to adapt to a performance measure a posteriori, i.e., a plug-in method. Surprisingly, little attention has been paid to this combination of a bagging ensemble and threshold-moving. In this paper, we study this combination and demonstrate its competitiveness. Contrary to the other resampling methods, we preserve the natural class distribution of the data resulting in well-calibrated posterior probabilities. Additionally, we extend the proposed method to handle multiclass data. We validated our method on binary and multiclass benchmark data sets by using both, decision trees and neural networks as base classifiers. We perform analyses that provide insights into the proposed method.

Business Analytics in Telemarketing: Cost-Sensitive Analysis of Bank Campaigns Using Artificial Neural Networks

“…The majority of literature considered re-sampling approaches, i.e., both over-sampling and under-sampling, to alleviate degradation due to the issue of imbalanced data [1,17,19,33,37]. Recent research contributions warn from the limitations and shortcomings accompany re-sampling approaches [16,38,39]. In particular, questionable reliability of produced models, i.e., while under-sampling approach may discard important instances, over-sampling approach may result in generating over-fitting models [30].…”

Boosting Minority Class Prediction on Imbalanced Point Cloud Data

“…Another strategy is the threshold-moving technique in which the decision threshold is shifted in a manner that reduces bias towards the negative class [15][16][17]26]. It applies to classifiers that, given an input tuple, return a continuous output value.…”

Bagged ensembles with tunable parameters

“…One such example is for the binary class imbalance problem, where Hido et al applied a variant of bagging (called roughly balanced bagging ), which uniformly samples the classes to mitigate the imbalance. As an extension to this, Lango and Stefanowski show that roughly balanced bagging can also be applied the multiclass imbalance setting . These examples show the practicality of not only bagging but also its conjunction with other difficult‐to‐solve issues.…”

“…As an extension to this, Lango and Stefanowski 15 show that roughly balanced bagging can also be applied the multiclass imbalance setting. 16 These examples show the practicality of not only bagging but also its conjunction with other difficult-to-solve issues.…”