Learning a fuzzy decision tree from uncertain data

Uncertainty in data exists when the value of a data item is not a precise value, but rather by an interval data with a probability distribution function, or a probability distribution of multiple values. Since there are intrinsic differences between uncertain and certain data, it is difficult to deal with uncertain data using traditional classification algorithms. Therefore, in this paper, we propose a fuzzy decision tree algorithm based on a classical ID3 algorithm, it integrates fuzzy set theory and ID3 to overcome the uncertain data classification problem. Besides, we propose a discretization algorithm that enables our proposed Fuzzy-ID3 algorithm to handle the interval data. Experimental results show that our Fuzzy-ID3 algorithm is a practical and robust solution to the problem of uncertain data classification and that it performs better than some of the existing algorithms.