Cost‐sensitive methods of constructing hierarchical classifiers

Abstract

Abstract:The cost of a future exploitation of a decision support system plays a key role. The paper deals with the problem of feature value acquisition cost for such systems. We present a modification of a cost-sensitive learning method for decision-tree induction with fixed attribute acquisition cost limit. Properties of the concept are established during computer experiments conducted on chosen benchmark databases from the UCI Machine Learning Repository and a real medical decision task. The results of experiments confirm that, for some decision problems, our proposition allows us to obtain a classifier with the same quality as a classifier obtained without cost limit but its exploitation is cheaper.

Evolutionary undersampling boosting for imbalanced classification of breast cancer malignancy

“…They usually combine an ensemble learning algorithm with one of the A c c e p t e d M a n u s c r i p t previous techniques, and more specifically, data level and cost-sensitive ones. In the case of data level approaches, a selected pre-processing algorithm is used before training each classifier of the ensemble to balance the class distribution [12,56,23].Cost-sensitive ensembles work on the basis of costs in the ensemble learning algorithm, cost-sensitive evaluation can be used locally for each base classifier [52] or globally as an evaluation metric [58].…”

Tracking a benchmark index – using a spreadsheet‐based decision support system as the driver

“…The decision support system (DSS), an interactive, flexible, and adaptable computer‐based information system with an easy‐to‐use interface, is designed specifically to support the solution of semi‐structured management problems and thereby improve decision making (Xie, ). Optimisation modules have also been used to support the solution of analytical problems (Turban, ; Karmakar et al ., ; Tsai, ; Goodwin & Russomanno, ; Kong et al ., ; Du et al ., ; Kuo, ; Penar & Wozniak, ; Chuang & Huang, ; Day & Nandi, ).…”

Empowering one-vs-one decomposition with ensemble learning for multi-class imbalanced data

“…• Cost-sensitive level: these approaches consider higher costs for misclas-115 sifying the minority classes with respect to the majority classes, that 116 is, misclassification of minority class is much more expensive [44] In the above reviewed methods, sampling process is independently carried 156 out before the training process. In these approaches, the training process is In order to predict a new pattern, there are two phases required to com-211 plete the task.…”