Partitionnement de données | EPFL Graph Search

vignette|upright=1.2|Exemple de clustering hiérarchique. Le partitionnement de données (ou data clustering en anglais) est une méthode en analyse des données. Elle vise à diviser un ensemble de données en différents « paquets » homogènes, en ce sens que les données de chaque sous-ensemble partagent des caractéristiques communes, qui correspondent le plus souvent à des critères de proximité (similarité informatique) que l'on définit en introduisant des mesures et classes de distance entre objets. Pour obtenir un bon partitionnement, il convient d'à la fois :

minimiser l'inertie intra-classe pour obtenir des grappes (cluster en anglais) les plus homogènes possibles ;
maximiser l'inertie inter-classe afin d'obtenir des sous-ensembles bien différenciés.

Vocabulaire La communauté scientifique francophone utilise différents termes pour désigner cette technique. Le mot anglais clustering est communément employé. On parle également souvent de méthodes de regroupement. On

Clustering of winter precipitations in Valais

Clémence Marie Anne Le Cornec

This master thesis provides an extended study of winter precipitations in Valais (Switzerland). It was rst tried to relate local conditions in Valais, as measured at gauges, to the Hess-Brezowski classication for Europe. It was not possible to clearly relate this classication to local precipitation events in Winter in this region. Self-organizing maps were used to try to distinguish between sub-regions only based on the precipitation regime in Valais. The obtained results present coherence with the existing literature. ERA-Interim (global atmospheric reanalysis) data were then used to develop a new classication to link European meteorological conditions to local events. Three unsupervised clustering techniques were used for this purpose: k-means, k-medoids, hierachical clustering. The optimal number of cluster was determined using a com- bination four commonly used internal clusters indices (Dunn index, Davies-Bouldin index, silhouette index and Calinski-Harabasz index). Two clusters appeared to be optimal compared to the various parameters of the problem. The outputs of the three classication techniques were compared in term of robustness and results. A physical interpretation was then made. It was found that the two classes were associated respectively with small or large precipitation events. The associated synoptic conditions at the European scale reveal that the small precipitation class is associated with a North-East ux over Switzerland whereas the large precipitation class is linked to a South-West ux linked to a stationary front over Switzerland.

2017

Systematic analysis of low-affinity transcription factor binding site clusters in vitro and in vivo establishes their functional relevance

Shiyu Cheng, Ivan Istomin, Maria Lopez Malo, Sebastian Maerkl, Evan James Olson, Amir Shahein

Binding to binding site clusters has yet to be characterized in depth, and the functional relevance of low-affinity clusters remains uncertain. We characterized transcription factor binding to low-affinity clusters in vitro and found that transcription factors can bind concurrently to overlapping sites, challenging the notion of binding exclusivity. Furthermore, small clusters with binding sites an order of magnitude lower in affinity give rise to high mean occupancies at physiologically-relevant transcription factor concentrations. To assess whether the observed in vitro occupancies translate to transcriptional activation in vivo, we tested low-affinity binding site clusters in a synthetic and native gene regulatory network in S. cerevisiae. In both systems, clusters of low-affinity binding sites generated transcriptional output comparable to single or even multiple consensus sites. This systematic characterization demonstrates that clusters of low-affinity binding sites achieve substantial occupancies, and that this occupancy can drive expression in eukaryotic promoters.

NATURE PORTFOLIO2022

Bases méthodologiques de la participation lors de projets ayant des impacts sur le paysage

Vincent Luyet

Recently, participation emerged as an appropriate tool to enhance natural resource management. Despite official recognition, there is no consensus on the practical meaning of participation. In this work, we consider the participation as a process through which stakeholders influence and share control over development initiatives and the decision and resources affecting them. The aim of this research is to answer to the participation's issues : who should participate? when? and how? The main contribution of this reserach work is to propose a methodology that allows to introduce the participation for large projects having impacts on the landscape. The result of this methodology is a system of participation. This system is composed of elements (project, stakeholders, techniques and levels of participation) and processes (identification, characterisation and structuration of the stakeholders, identification of the participation level, choice of the participation techniques, application and evaluation of the participation) in interaction. This system of participation is general and can be used for others projects. Howewer, the chronological order of the processes is iterative and is project specific. For each of the process used in the participation system, techniques were adapted, deveopped and tested. Stakeholders identification followed the adapatation of the Mason and Mitroff technique. Stakeholders structuration was made using cluster analysis. Identification of the level of participation followed the modifications of the Daniels et al and Vroom models. Relevent identification and assessment of participation techniques were made with the help of tables. All these thechniques yield results readily and pratically usable. The way stakeholders perceive the landscape and its dynamics are also studied in this research. This study allows to analyze the type and the effects of the information presented on the participative process and on the stakeholders. Based on our results, we advise to present landscape dynamics in a participative context. We recommand the use of simple landscape metrics such as the area doubled with the spatialization of the transition matrix. The stakeholders landscape preference allows to identify criteria such as the attitude towards the changes that would help in the elaboration of the future project. The originalities of this study can be found at various levels. In a general way, we can list the globality of the proposed methodology, the overlap of the social and technical sciences and the application of the methodology to the third Rhone correction project. In a more specific way, we can mention technical developments such as the application of original experiments concerning the identification of stakeholders landscape preferences. The system of participation proposed provides a conceptual framework that can be applied to others participative projects such as agenda 21 or the development of any projects having impacts on the landscape.

EPFL2005

Bases méthodologiques de la participation lors de projets ayant des impacts sur le paysage

Vincent Luyet

EPFL2005