Cost-Sensitive Classification: Empirical Evaluation of a Hybrid Genetic Decision Tree Induction Algorithm

By Peter D. Turney. 1994

This paper introduces ICET, a new algorithm for cost-sensitive classification. ICET uses a genetic algorithm to evolve a population of biases for a decision tree induction algorithm. The fitness function of the genetic algorithm is the average cost of classification when using the decision tree, including both the costs of tests (features, measurements) and the costs of classification errors. ICET is compared here with three other algorithms for cost-sensitive classification -- EG2, CS-ID3, and IDX -- and also with C4.5, which classifies without regard to cost. The five algorithms are evaluated empirically on five real-world medical datasets. Three sets of experiments are performed. The first set examines the baseline performance of the five algorithms on the five datasets and establishes that ICET performs significantly better than its competitors. The second set tests the robustness of ICET under a variety of conditions and shows that ICET maintains its advantage. The third set looks at ICET's search in bias space and discovers a way to improve the search.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Mathematical optimization in classification and regression trees (2021, background): “The fairness can refer to the accuracy achieved in risk groups, for which the consequences of a wrong prediction are much more severe than for the rest (Kao and Tang 2014; Turney 1995).”
Strongly agree or strongly disagree?: Rating features in Support Vector Machines (2016, background): “One may also like to keep the costs of learning the classi er low, where these costs arise from retrieving the features (Carrizosa et al. 2008, Turney 1995), obtaining feature information from the classi er (Saar-Tsechansky et al. 2009), or refreshing the classi er to include new data (Fang et al.…”
A systematic study of the class imbalance problem in convolutional neural networks (2017, methodology): “For example, cost curves [100, 101], regret graphs introduced by Hilden and Glasziou [102], performance graphs used by Turney [103], Loss Difference Plots, and LC index [104].”
A Survey of Evolutionary Algorithms for Decision-Tree Induction (2012, background): “to improve tree pruning [17], [18] and other components that are related to decision-tree induction [19], [20].”
Bayesian Artificial Intelligence (2004, unknown): “For an example of recent work on cost-sensitive classification, see papers by Turney [276] and Provost [224]; there was also some early work by Pearl [214].”
Meta-learning by the Baldwin effect (2018, methodology): “Turney used a genetic algorithm to evolve a population of biases for a decision tree induction algorithm for classiﬁcation [28].”
Cost-based modeling for fraud and intrusion detection: results from the JAM project (2000, background): “According to Turney [28] the earliest work here is due to Nunez [17].”
Finding Cost-Efficient Decision Trees (2014, methodology): “The cost heuristics considered when designing my algorithm are adopted from Turney’s[42] cost-sensitive classification of decision trees.”
Supervised classification and mathematical optimization (2013, background): “Therefore, it may be desirable to assign costs to features to discriminate among them [227].”
Types of Cost in Inductive Concept Learning (2002, methodology): “When an induced model is used to gain understanding of the underlying process that generated the data, it is important that the model should be stable (Turney, 1995c; Domingos, 1998).”
Real Time Credit Card Fraud Detection using Computational Intelligence (2007, background): “It is necessary to employ an effective and economical solution to combat fraud (Turney, 1995).”
Connectionist Theory Refinement: Genetically Searching the Space of Network Topologies (1997, unknown): “GAs have been combined with many local search methods (Bala, Huang, Vafaie, DeJong, & Wechsler, 1995; Belew, 1990; Hinton & Nowlan, 1987; Turney, 1995).”
Hybrid decision tree and naïve Bayes classifiers for multi-class classification tasks (2014, background): “The goal of DT is to create a model that predicts the value of a target class for an unseen test instance based on several input features (Loh & Shih, 1997; Safavian & Landgrebe, 1991; Turney, 1995).”
Multi-Objective Particle Swarm Optimization Approach for Cost-Based Feature Selection in Classification (2017, background): “may come from economy, time, or other resources used to obtain feature values of objects [9], [10].”