Abstract: A new, lossy data reduction method based on information theory is described in this paper, which overcomes problems associated with other data reduction methods. It is shown that the effectiveness of this method can handle complex datasets without losing information trends.