Decision Tree is one of the widely used data classification techniques. This paper proposes uncertain decision tree classification method. Lots of Factors causes Value uncertainty including measurements precision limitation, outdated sources, lack of information, and transmission problems. With uncertainty, the value of a data item is often represented not only by one single value, but also by multiple values forming a probability distribution.

Data of family economical information plan survey are uncertain because of reticence and lack of data. We need to have appropriate algorithm to work with uncertain data with satisfactory accuracy. In this paper, we upgrade the traditional uncertain decision tree algorithm, using entropy and information gain, and extend measures, including the uncertain data interval and probability distribution function which help in reducing the demanding effects of imbalance data on the output of algorithm. Our algorithm can handle both certain and uncertain datasets. This paper indicates that, the proposed algorithm has satisfactory prediction accuracy.

Uncertain Decision tree construction on data use much more CPU than that for certain data. To tackle this problem, we propose a max level technique that can greatly improve construction efficiency.