Dresden 2009 – scientific programme

Parts | Days | Selection | Search | Downloads | Help

SYCS: Symposium Data Analysis in Complex Systems

SYCS 1: Data Analysis in Complex Systems: From Data to Models

SYCS 1.5: Invited Talk

Friday, March 27, 2009, 12:30–13:00, BAR SCHÖ

Clustering, chance, and statistical mechanics — Marta Luksza¹, Michael Lässig², and •Johannes Berg³ — ¹Max Planck Institut für molekulare Genetik, Berlin — ²Institut für Theoretische Physik, Universität zu Köln — ³Physikalisches Institut, Albert-Ludwigs-Universität Freiburg

Clustering a set of points in a high-dimensional space is a core method of statistical data analysis. We consider the problem of clustering in random data: Given a set of randomly distributed vectors, how likely do some of them form a cluster with a given cluster score? This cluster p-value is crucial to assess the statistical significance of clusters found in real data. We use a mapping to a disordered spin-model to analytically solve the random clustering problem and discuss applications to the analysis of gene expression patterns.