Mathematicians Find Better Way to Reliable Predictions

March 26, 2008

Brown University mathematicians have developed of a class of statistical estimators that could lead to better methods of analyzing large amounts of data.

Professor Charles Lawrence and graduate student Luis Carvalho concentrated on “centroid” estimators as the key to statistical predictions and better ways to extract information from immense data sets found in computational biology, information technology, banking, finance, medicine, and engineering. These centroid estimators, instead of identifying the single most probable estimate, point to a solution that is most representative of the data in any large set.

The mathematicians used statistical decision theory to anticipate the need for the most common estimation procedure--maximum likelihood estimation--when faced with "high-dimensional" inference problems. Then they used statistical decision-making theory to find an estimation procedure for a broad range of statistical problems. Finally, Lawrence and Carvahlo proved a number of theorems to highlight the favorable properties of these estimators, and demonstrated that they could actually be computed in numerous applications.

“What’s exciting about this work, what makes it every scientist’s dream, is that it’s so fundamental,” Lawrence said. “These new estimators have applications in biology and beyond, and they advance a statistical method that’s been around for decades.”

Their paper, "Centroid Estimation in Discrete High-Dimensional Spaces with Applications in Biology," was published in Proceedings of the National Academy of Sciences (2/27/08).

Source: Brown University