This study uses cluster analysis as the methodology to explore policy of the asset allocation as well as the selection of equities under the multiple-factor asset pricing models. Based on the data of financial market recorded on Bloomberg from 2000/1/4 to 2012/2/10, the conclusions of this study are summarized as following: First at all, under the significance level of 5%, P/S ratio should be included in the multiple-factor asset pricing model. Nonetheless, the significance of proxy agent of foreign exchange volatility in terms of 11-day moving average of USD/JPY foreign exchange spot rate, as well as the interest spread in terms of yields on 10-year US government bond subtracting 3-month US treasury bill cannot pass the required significance level. Second, the rates of stock return as Qualcomm, Intel and Texas instruments in the industry supply chain of technology products, will be positively related to interest spread, with the variable of ¡§Foreign_Volitility¡¨ negatively related to those rates of return as well as sales growth momentum positively related to those. As far as those rates of stock return 3C brand companies such as Apple, Microsoft, Dell and IBM, the interpreting capability of variable of ¡§Foreign_Volitility¡¨ under the assumptions of market structure in this study, will be mixed, with the interest spread positively related to those returns and P/S ratio generating mixed outcomes. As far as those equities such as GE, Procter & Gamble, Home Depot, Tiffany, AIG, NIKE, Exxon Mobile Corp, the interpreting capability of variable of ¡§Foreign_Volitility¡¨ under the assumptions of market structure in this study, will be negatively related to stock return except for Exxon Mobile Corp, with the interest spread generating mixed outcomes and P/S ratio positively related to those returns.

Les modèles cinématiques de plissement permettent de prévoir l'architecture des anticlinaux en profondeur à partir de leurs géométries de surface. Bien que les modèles classiques de type « kink-band » soient faciles à utiliser, ils ne peuvent expliquer que des géométries de plis limitées. Le modèle trishear surmonte cet inconvénient, mais son application reste difficile à mettre en oeuvre notamment en raison de nombreux paramètres mal contraints tels que le rapport P/S et l'angle de trishear. Dans cette thèse, nous avons développé un modèle cinématique de plissement à angle d'inter-flancs variable qui permet de rendre compte de toutes les variétés géométriques de plis sur propagation de faille. Ce modèle est basé sur la géométrie « kink-band » et son application nécessite de connaître uniquement les paramètres de surface du pli (plongements des flancs, amplitude et longueur d'onde du pli). Le modèle permet de quantifier le rapport P/S du pli durant sa croissance, ainsi que son évolution tardive en pli de rupture. Le modèle a été appliqué à la chaîne plissée et faillée de l'Atlas tunisien pour estimer la profondeur du niveau de décollement des plis sur propagation de faille. Les résultats se corrèlent bien avec les données de subsurface, au contraire des autres modèles classiques et du modèle de trishear où il est difficile d’établir une relation entre la géométrie de surface des anticlinaux et leur architecture profonde. / Kinematic models of folding allow deducing the deep architecture of folds from their surface geometries. Classical models are easy to use because of their kink-band geometry but they are only able to explain some particular fold shapes. The trishear model overcomes this drawback, but its application remains limited because it necessitates the knowledge of several unconstrained parameters such as the P/S ratio and the trishear angle. In this work, we developed a kinematic model based on the variation of the inter-limb angle during the folding. The model allows accounting for all surface geometries of fault-propagation folds. Since this model is based on the kink-band geometry, its application only requires the knowledge of the fold surface parameters, i.e. the dip of both limbs and the fold amplitude and wavelength. The model is also able to quantify the P/S ratio during the fold growth, together with its latest evolutional stage as a breakthrough fold. We applied the model in the Tunisian Atlas fold-and-thrust belt to estimate the depth of the decollement level associated to fault-propagation anticlines. Results are in good agreement with subsurface data. In contrast, both the other classical and trishear models are unable to explain the relationship between the surface geometry of anticlines and their deep architecture.