Dealing with data
Harvard researchers have developed a tool for analyzing large data sets that detects important relationships in data without prior knowledge of their type. The development comes at a time when researchers are being overwhelmed by the vast amounts of data emerging from their labs, and struggling to make sense of them. Developed by brothers David Reshef, a current Harvard-MIT M.D./Ph.D. student, and Yakir Reshef ’09, together with Professors Michael Mitzenmacher and Pardis Sabeti of the Harvard Departments of Computer Science and Organismic and Evolutionary Biology, respectively, the data-analysis algorithm is capable of quickly analyzing massive data sets to identify variables that may be related, enabling researchers to pick out potentially meaningful results they might otherwise have missed. The paper describing the algorithm, published in the Dec. 16 issue of Science, applies the program to four data sets — microbiome data, genetic studies, global health data, and baseball statistics — in an effort...
