Adaptive Cost-sensitive Decision Tree Algorithm And Its Application

With the development of database technology and information technology, Human has accumulated a large amount of data. In order to dig out new valuable effective information from this large amount of data and meet the needs of users, data mining technology developed rapidly, and recently has been concerned greatly in some areas. As a commonly used data mining technique, classification technology continues to be widely used in all walks of life. Among all the classification algorithms, because their model structures are simple and intuitive, and are easy to be understood and interpreted, and can effectively solve many practical problems, the classification algorithms of decision tree achieve rapid development.This paper first introduces the model of decision tree and some classical decision tree classification algorithms, and by comparing the advantages and disadvantages of these algorithms, proposes an improved adaptive cost-sensitive decision tree algorithm. Then based on a set of medical diagnostic dataset, we create a cost-saving tree with the improved decision tree algorithm, and comparatively analysis the performance of the improved algorithm, and find that the improved algorithm has a lower average cost of classification and good robustness under good accuracy. Thus this article can provide a useful reference for the decision tree model used to analysis how to save the cost of the diagnosis of a medical disease, and so has some practical significance.