Objectifs et indicateurs SMART

vignette|Acronyme anglais du terme SMART. SMART (intelligent en anglais) est un moyen mnémotechnique permettant de décrire les objectifs que l’on veut exprimer de façon la plus claire, la plus simple à comprendre et pour lesquels les résultats sont réalisables. Un indicateur est une variable permettant de fournir des informations pour chacune des étapes d’un projet afin d’aider à la bonne prise de décision. L’analyse des différents indicateurs retenus se fait selon différents critères tels que la quantité des informations disponibles pour une étude, les coûts engendrés par la mise en place du projet et les ressources que l’on peut en obtenir, etc. Le concept de management par objectifs est défini en 1954 par Peter F. Drucker sans employer l’acronyme SMART. La méthode consiste à identifier des objectifs quantitatifs et/ou qualitatifs sur une période définie. Il ajoute également que pour calculer la performance, il faut prendre en compte les travailleurs dans la détermination des objec

A tutorial on adaptive MCMC

Johannes Thoms

We review adaptive Markov chain Monte Carlo algorithms (MCMC) as a mean to optimise their performance. Using simple toy examples we review their theoretical underpinnings, and in particular show why adaptive MCMC algorithms might fail when some fundamental properties are not satisfied. This leads to guidelines concerning the design of correct algorithms. We then review criteria and the useful framework of stochastic approximation, which allows one to systematically optimise generally used criteria, but also analyse the properties of adaptive MCMC algorithms. We then propose a series of novel adaptive algorithms which prove to be robust and reliable in practice. These algorithms are applied to artificial and high dimensional scenarios, but also to the classic mine disaster dataset inference problem.

2008

Evaluation de stratégies pour la gestion du risque sismique du bâtiment

Vincent Pellissier

This document presents the results of a research on seismic risk management. It echoes the worldwide trend of increasing awareness of the importance of this risk and represents a contribution to the effort of research. In order to manage risk in a rational way, it is necessary to quantify it. The first part of this research consists in evaluating the seismic risk of a portfolio of buildings. The selected portfolio is the city of Aigle located in the center of the Swiss Alps. The evaluation was based on a seismic inventory of the city, enabling, through a global approach, to consider the behavior of the buildings. Different losses (life, building, furniture) could be estimated for several scenarii of seismic hazards intensities. Knowing the probability of occurency of these intensities, as well as the uncertainties of the constitutive parameters, it was then possible to quantify the risk affecting the selected portfolio of buildings. The maximum potential loss for the studied area was also estimated. As a result, a reproducible risk estimation model was developed with its constitutive parameters. The obtained results provide information on potential losses, on the level of insurance premium as well as on elements of reinsurance. Seismic risk management, because of its complexity and the extent of the consequences ensuing from major events, requires a different approach from other natural risks. In the second part of this research, management tools are therefore developed in order for decision makers to face these particularities. A comprehensive management framework is elaborated. This framework makes it possible to deal with the complexity of seismic risk, which not only stems from its inherent specificities but also from the multiplicity of decision criteria as well as the variety of actors within the system. The definition of protection objectives receives special attention in this research. Construction codes usually fix safety levels for new constructions. For existing buildings on the other hand, there are no such provisions. The present research proposes a method for fixing this protection level. This method also considers the often contradictory principles of precaution and proportionality. An innovative catalogue of mitigation measures are then proposed, classified according to their influence on the components of risk. These measures can be focused on prevention, protection or intervention. They have an impact on the hazard, the vulnerability or on the exposed value at the considered risk. Consequently, different strategies are built to be later evaluated and compared. For this work, a strategy is considered to be a panel of measures that have an impact on the constitutive elements of risk. The decision-making process is regarded as a systemic approach considering the multiplicity of decision criteria and actors. A multicriteria analysis method based on an outranking philosophy is adopted for the comparison. Criteria are proposed for individual buildings as well for a portfolio of buildings. Applications of the decision-making process on pilot projects illustrate the methodology and provide practical recommendations. The pilot project of the city of Aigle was the guideline for the presented research, which aimed to deal with the complexity of seismic risk management in the most comprehensive way possible.

EPFL2004

An Extensive Group Decision Methodology for Alliance Partner Selection Problem in Collaborative Networked Organisations

Omar Ayadi, Naoufel Cheikhrouhou

More and more companies address collaborations and cooperation as strategic topics to build competitive networks and broaden the global competences provided by the community thus created. These companies can cooperate with partners to share resources, competences, risks, or costs. Besides, penetrating new markets can be easier when associating new partners. The selection process of an adequate partner considered for a specific objective is a key success factor. This paper proposes an evaluation methodology for selecting an alliance partner in the case of a network of enterprises, manufacturing high precision mechanical components. An extensive group decision methodology is developed using both quantitative data and qualitative judgments in the evaluation of criteria. Since some performances cannot be represented with crisp numbers, the proposed methodology allows experts to use linguistic variables to express their judgments for the assessment of qualitative criteria. Two main phases are integrated in this methodology. In the first phase, criteria and experts weightings are calculated to determine the criteria importance using fuzzy the analytical hierarchy process. In the second phase, a technique for order preference by similarity to ideal solution is proposed to rank the different alternatives, corresponding to the alternative partnerships.

2012