In many applications, an important fact is that the stream data may not show change on all sets of the dimensions, but on some particular combinations in which dimensions are correlated with each other. Therefore, it is important to develop tools and techniques to provide a diagnostic overview of correlations in evolving data streams in a fast and user-friendly way. In this paper, we provide a system for mining linear correlations and lag correlations among data streams. In our system, we provide capturing and tracking changes of linear correlations in terms of stream data evolution. We also aim to discover pairs of data streams correlated with lags, as well as the value of each such lags. Experimental results illustrate that our correlation mining mechanism performs effective ly and efficiently in prediction of stream data.