Lespinats, S., Giron, A. et Fertil, B. (2005) “Visualisation and exploration of high-dimensional data using a "force directed placement" method: application to the analysis of genomic signature.” International Symposium on Applied Stochastic Models and Data Analysis, ASMDA 2005, mai 2005, pp. 230-237.
Visualization of high-dimensional data is generally achieved by projection in a low dimensional space (usually 2 to 3 dimensions). Visualization is designed to facilitate the understanding of data sets by preserving some "essential" information. We have designed a non-linear multi-dimensional-scaling (MDS) tool relying on the force directed placement (FDP) paradigm to help dynamically discover features of interest in data sets. A user-driven relaxation of constraints built on the preservation of pairwise distances between data allows getting subjective representations of data that meet some specific angle. In a context of classification, we examine the impact of metric, sample size, and neighborhood preservation on the mapping of genomic signatures.Télécharger l'article Télécharger la présentation