Cost-Sensitive Decision Tree Induction on Dirty Data

As the rapid growth of data in our society, dirty data are increasingly common. In the process of cost-sensitive decision tree induction, dirty data in training data sets have negative impacts on the selection of splitting attributes and division of decision tree nodes. Hence, dirty data cleaning is necessary before classification tasks. However, many users give an acceptable threshold of data cleaning costs since time costs and expenses of data cleaning are expensive. Therefore, in addition to misclassification cost and test cost, data cleaning cost is an important factor in cost-sensitive decision tree induction. However, existing researches have not considered data quality issues in the problem. To fill this gap, this chapter aims to focus on cost-sensitive decision tree induction on dirty data and presents three decision tree induction methods integrated with data cleaning algorithms. Evaluation results demonstrate the effective of the proposed approaches. Section 7.1 gives the research background. Section 7.2 introduces the problem definitions of this chapter. Section 7.3 discusses three proposed cost-sensitive decision tree building methods. Section 7.4 analyzes the experiments and evaluation results. Section 7.5 concludes the chapter and provides some future research directions.