Estimating probability values from an incomplete dataset

An essential component in Machine Learning processes is to estimate any unc ertainty measure reflecting the strength of the relationships between varia bles in a dataset. In this paper we focus on those particular situations wh ere the dataset has incomplete entries, as most real-life datasets have. We present a new approach to tackle this problem. The basic idea is to initia lly estimate a set of probability intervals that will be used to complete t he missing values. Then, these values are used to obtain new bounds of the expected number of entries in the dataset, The probability intervals are na rrowed iteratively until convergence, We have shown that the same processes can be used to estimate both, probability intervals and probability distri butions, and give conditions that guarantee that the estimator is the corre ct one. (C) 2001 Elsevier Science Inc. All rights reserved.