DATA ANALYSIS

Author: MARCOS CALVO MIGUEL ÁNGEL.
Year: 2003.
University: REY JUAN CARLOS.
Place of defense: CIENCIAS JURÍDICAS Y SOCIALES.
Place of preparation: FACULTAD CIENCIAS JURÍDICAS Y SOCIALES.

Summary: Despite the economic and social cohesion between regions is a simple definition of politics is not so much from a technical point of view, ie is a very complex to assess and / or quantify. This thesis proposes a methodology for quantifying on the idea that economic and social cohesion inferred the existence of inequality between regions admissible considering jointly and simultaneously both economic and social factors. Indirectly, it is proposing to overcome the customary use of Capita GDP PER as the sole indicator for analysis and decision in regional policy. The thesis distinguishes two parts: an initial concept which provides far from the theoretical point of view, stressing the basis of economic theory, political, and economic statistics needed for its definition, and a second thumb, which contrasts his aplicaicón Case the European regions. The statistical basis of the measure lies in the implementation of a set of methods belonging to the discrete multi decision, known as methods electre, but used in a manner close to the data analysis. Specifically, they get sort and classify a whole regions with a view to jointly and simultaneously a group of socio-economic factors. In conclusion and seeking a broader applicability, a criterion for eligibility in multiple variables directly observable. Benchmark understood as a socio-economic profile of classification so that any region to submit a budget to lower the profile is in a position of inequality unacceptable with respect to other regions.

Author: GONZÁLEZ PÉREZ BEATRIZ.
Year: 2004.
University: COMPLUTENSE DE MADRID.
Place of defense: FACULTAD.
Place of preparation: FACULTAD DE MATEMÁTICAS.

Summary: It develops a Bayesian procedure for the contrast of homogeneous populations multinomiales independent contingency tables rxs, when the vector of common shares in the null hypothesis is known, unknown or functionally known. The methodology is to assign a mass to the initial ante null hypothesis and spread the probabiidad remaining points of the alternative through a density function. It is estimated the probability post the null hypothesis of homogeneity for, and compared with the p-valor classic usual procedure of obtaining a reconciliation between the two methods in terms of a sufficient condition. Finally, in the context of the null hypothesis multivariate timely, it demonstrates a theorem that shows you when you and how it is possible to reach an agreement between the classical and Bayesian approaches.

Author: LLORENTE GALERA FRANCISCO.
Year: 2004.
University: BARCELONA.
Place of defense: FACULTAD DE CIENCIAS ECONÓMICAS Y EMPRESARIALES.
Place of preparation: UNIVERSIDAD DE BARCELONA.

Summary: We analyzed within the main strategies to improve the competitiveness of the automotive sector innovation. Specifically discusses the technological and organizational innovations that are being introduced in the sector, as if taking significant aSEAT. It has sought to know the developments in technology products, stressing the incorporation of electronics and the variety of products it offers. It reflects the importance of technology innovations in flexible manufacturing processes and virtual simulation. It says the survey of the INE technological innovation of the year 2000 concerning Spain and Catalonia. In the case of SEAT describes the functions of its Technical Center performs and how the respective R & D process at various stages. In organizational innovation reflects the new relationship manufacturer / supplier and labor practices of the new labor organization. In part, we have made various empirical surveys to know the reality of the direct suppliers located in Catalonia automakers and manufacturers in Spain, in terms of incorporating organizational innovation has been an impact on the potential level of introduction of certain practices labor associated with the lean production. We have found that organizational systems that are implanted companies mechanistic hybrid between the organization and the new organization work. While some items are incorporated so important, we see a small incorporation joint high analyzed practices associated with the new labor organization. In suppliers has also interested to know whether conducting technological innovation, if they use certain techniques and methods that help reduce time to market, as well as incorporate the strategy of collaborating with external agents. This produces a small joint integration of the systems, techniques and methods for managing innovation. The collaboration with customers and suppliers is high, but not when considering the overall external agents. It has differentiated between companies in domestic and foreign capital, by size and sector of business ownership. The exploitation of such statistical surveys have been used univariate analysis techniques, bivariate and multivariate (Homals and Cluster), as well as realiuzación contrast non - parametric.

Author: CANITO LOBO JOSÉ LUIS.
Year: 2004.
University: EXTREMADURA.
Place of defense: ESCUELA DE ING. INDUSTRIALES.
Place of preparation: ESCUELA DE ING. INDUSTRIALES.

Summary: The towns are provided with services that are cataloged in infrastructure and equipment, and depend on several factors. There are 14 individuals who reflect the state of these services. Each of these indicators, which are expressed in different units ediola reflect the status of each of the towns. This also creates the possibility of obtaining one ranking from discriminating against various towns and reflecting the peculiarities of the individual rankings. The aim has been designed in a -- that allows us to the theoretical development of a theoretical ---, --- in terms of a -- of the indices that we discriminate in unison, both nuclei, as indices .

Author: RUANO PANIAGUA NOELIA MARÍA.
Year: 2004.
University: POLITÉCNICA DE MADRID.
Place of defense: E.T.S. ING. CAMINOS, CANALES Y PUERTOS.
Place of preparation: E.T.S. ING. DE CAMINOS, CANALES Y PUERTOS.

Summary: The term fatigue refers to the process of progressive and irreversible deterioration experienced when a material is subjected to cyclic stresses and repeated below even the elastic limit of the same. Often described by a parameter is called fatigue limit which represents the amplitude voltage threshold below which it is considered that the material is not going to fail for fatigue even though the number of cycles is very high. The breadth of tension is usually the variable that most influences the fatigue so that the larger, the lower the number of cycles that aguantará material. Fatigue in concrete structures ever created as much interest as those of steel: whereas in the latter fatigue has caused catastrophic damage in the first, there have been no such rulings, although in some cases there has been speculation about whether the behavior real could have been attributed to fatigue. The investigations carried out in the concrete have indicated that the effect of repeated loads can be more harmful than has been thought, because the concrete structures are becoming increasingly slim, and the trend is directed towards better utilization components of the materials, particularly steel. For example, the United States bars corrugated fatigue has not been a factor in its application as reinforcement in concrete structures. However, in recent years it has developed a strong interest in the study of the phenomenon of fatigue armed elements. The reasons were as follows: - The widespread adoption of design procedures based on the last resistance and the use of materials with higher benefits, require concrete structural elements behave satisfactorily compared with high amplitudes of tension . Hence the importance that acquires the effect of repeated charges in crane and slabs of bridges. - Increased use of concrete in marine environments, aggressive and repeated loads, rail structures such as bridges, rail beams and other applications, it is necessary to make high performance to ensure a certain resistance to fatigue. Currently being developed new or different uses of the elements of concrete as strained railway prestressing concrete or reinforced concrete pavements continuing. - We know that the actions of the repeated loads can change the width of the cracks, deformation and the stiffness of the structure in the States Limit Service. And even in some cases, produce a cracking in the material components enough to alter the load-bearing capacity compared to static loads. In the beams pretensadas, for example, it has been found that on some occasions there has developed a cracking angled for statements of charges lower than expected. The steel bars are limited elastic increasingly high. This, coupled with the fact that they have been placed at the point design methods that allow greater utilization and capacity utilization of materials of the structure, and the introduction of load factors, has led to an increase in the amplitudes tension repeated that at times are very close to breaking that cause fatigue steel bars. Notwithstanding the foregoing, have not yet signed any fatigue fracture of a steel reinforced concrete structure under burdens of service. However, by fatigue cracks in the armed due to the overloading of the bridges has drawn attention to drawing up the design specifications of the same. In addition, there is also a growing awareness towards the phenomenon of fatigue because of the experience you have regarding the operation of certain types of structures in the states Limit Service, such as foundations or offshore platforms ace 8 ro solda 1ff8 das. In light of the enumerated above, it is necessary to ensure the proper conduct to fatigue in a concrete structure, which the designer knows the solicitations capable of causing fatigue, the types of structures susceptible to the action of fatigue, and the minimum requirements due to the concrete and steel on the market, mainly in the conditions of service of the structure, so that they can cope successfully with fatigue. In the particular case of steel bars, his behavior has been investigated fatigue in qualitative terms: the impact of features such as the type of test, the type of bar, the influence of the extent of strain and stress media, etc., a trend that continues today for the dual need for inclusion in the new codes, on the one hand, information about the behavior of concrete structures subjected to fatigue and rules allowing for a safe design of the same and secondly, the study of materials that are part of them, essentially, steel and concrete. However, many of the available data only pertain to tests conducted under laboratory conditions that do not simulate the real effects of structural element, and others have been obtained for unusually high tension, or to numbers of cycles that do not relate to the life design of structures. But there is a loss of quantitative information about the characteristics that influence the fatigue of steel, for example, we know that the fatigue behavior of a bar corrugated decreases with the increase in diameter, but even today there is no basis pilot sufficiently broad to allow numerically express this relationship; different investigations have come to very different results, mainly due to the fact that the fatigue behavior of a bar is heavily dependent on aspects that are beyond the control of the investigator. This joins the fact that there may be, and indeed must, in the sparse pilot tests with the same material, or large differences between different bars coladas or manufactured with the same roller rolling but with a different degree of wear. All of this has led us to think that there is the practical need to tackle an experimental work that allows characterize the fatigue behavior of the bars corrugated manufactured in Spain in the quality B500SD and diameter 25 mm (as the material, according with literature, presents some characteristics of resistance to fatigue than the quality B400SD, but worse than other smaller diameters), and to develop a procedure for estimating statistical, based on a certain number of test results, curve WÃ ¶ hler of this steel. Because we must not forget that codes of concrete when referring to the State Limit Fatigue in a concrete structure, a number of characteristics (type amplitude tensión-número cycles) they have to be performed separately from the concrete and steel. In this regard, the objectives of this thesis are designed in a twofold: firstly, to assist in its task of providing designer to design an expression analytical curve WÃ ¶ hler so that it can be used in the calculation and it allows better use and more precise knowledge of the steel structure, taking into account the actions which will be subjected. On the other hand, by allowing the manufacturer to deepen the fatigue behavior of the material that is manufacturing, as well as the study of the causes that could lead to the initiation and propagation of fatigue cracks (inclusions, residual stress, etc.). To to introduce the necessary changes in the manufacturing process to help improve product quality. The experimental research has consisted of 5 stages. In the first category were the results of tests of fatigue depending on the area of the bar where there has been breakage. It should be borne in mind that the most important problem that arises in the trials fatigue is the frequent occurrence of failures jaws. They provide much less information than other statistics have been completed until the actual failure by fatigue because of the first the only thing that is known is that the number of cycles that bear a test tube (subjected to a certain extent voltage) higher than that for which occurred on the fault in the jaws. The second has been assumed that the random variable number of cycles to failure resisted, N, for each level of the extent of tension, followed by a distribution belonging to a particular family or type (normal logarítmico-normal, weibull, etc..). The third has been proposed various forms of analytical curve WÃ ¶ hler from the information contained in the literature, were raised different models, have discussed how to estimate the parameters of these models have been analyzed and the results presented and differences between the factories. The model chosen for the statistical estimation of the curve WÃ ¶ hler must have a correct treatment of the data censored (failures jaws and run-outs), which is essential in the statistical analysis of data from fatigue, providing the most credible the parameters of the model and to quantify the uncertainty in the estimates obtained. Once it is estimated curve WÃ ¶ hler for the various factories of the study and has been chosen as the representative that corresponds to the lower casing of all factories, in the fourth phase are estimated curves WÃ ¶ hler characteristics guaranteed steel quality B500SD and diameter 25 mm, depending on certain levels of security, and trust. However, the problem of obtaining characteristic values (or tolerance limits) unilateral, for certain, and for equations estimated by the method of maximum likelihood in the presence of censored data, it has not been addressed, to our knowledge, until date, it has been necessary to develop a solution own original character to obtain curves WÃ ¶ hler characteristics. Lastly, since they have been estimated values of the fatigue limit characteristic of steel and has proposed an alternative method for obtaining it in certain situations. The main conclusions to be drawn from this investigation and that has allowed the estimation of curves WÃ ¶ hler features from a set of tests that are fatigue, the models that have been tanteado to estimate curves, the best everyone has adjusted to the test results has been the Model 1 which provides a linear relationship between the median log (N), and a ditribución lognormal for N. Later this model has particularized adjusting to the results of each of the 8 factories that have entered the study, constantándose that there are statistically significant differences between them, and therefore it is not correct from the standpoint of statistical analysis, considering that all of the 192 test results belong to a single population, but rather to 8 different populations. But in relation log (N), the outstanding differences between the factories have been found are not statistically significant, and may portray themselves as straight parelelas that differ only in the value of orderly at the source. The curve corresponding to the Factory 2 is the one that, for the same value of N, the lowest. For this reason it has been chosen as the envelope curve bottom of all the factories, as guaranteed curve of WÃ ¶ hler of them all, already 8 s effect 6d3 s design, as a basis for obtaining it from the characteristic curve. In order to take into account the curve WÃ ¶ hler uncertainty associated with the imprecision in estimating the model parameters, it has introduced the concept of value or characteristic tolerance limit, associated with certain levels of security, and confidence. Because the problem of obtaining tolerance limits for unilateral equations estimated by the method of maximum likelihood in the presence of censored data has not been addressed so far, it has been necessary to develop a solution own original finding an expression approximate limit lower tolerance in which it has taken into account the presence of data censored and has been regarded as a simple linear regression model. In addition to the curve WÃ ¶ hler feature for certain, and we have obtained the limits of fatigue, depending on the number of courses taken, and has proposed a procedimieto alternative estimate.

Author: VALENCIA GARCÍA OLGA.
Year: 2005.
University: LEÓN.
Place of defense: FACULTAD DE CIENCINAS ECONÓMICAS Y EMPRESARIALES.
Place of preparation: FACULTAD DE CIENCIAS ECONÓMICAS Y EMPRESARIALES.

Author: CASTRO LÓPEZ MARÍA ÁNGELES.
Year: 2005.
University: ALICANTE.
Place of defense: ESCUELA POLITÉCNICA SUPERIOR.
Place of preparation: ESCUELA POLITÉCNICA SUPERIOR.

Summary: THESIS ON THIS HAVE BEEN PROPOSED NEW CRITERIA FOR SELECTION OF MODELS FOR MODELS REGRESIÓN LINEAR. THE MOST OF THESE ARE BASED ON THE CRITERIA DIVERGENCE ON JEFFREYS. BY ONE PARTY HAS BEEN DEFINED CRITERIA KICu AS AN ALTERNATIVE TO THE CORRECTION SEÑAL-RUIDO OF CRITERION KICc PROPOSED BY CAVANAUGH. THE OTHER PARTY, HAS BEEN DEFINED COMPLEXITIES OF A NEW MEASURE FOR PARENTS SIMÉTRICAS And DEFINIDAS POSITIVE, BASED ON J-DIVERGENCIA, WHICH HAS ALLOWED ESTABLISHING A NEW FAMILY APPROACHES, CRITERIA JCOMP. IN ADDITION TO MODELS REGRESIÓN LINEAR MULTIPLE AND MODELS REGRESIÓN MULTIVARIANTE, HAVE BEEN PROPOSED CORRECTED APPROACHES TO QUALIFY ICOMP, OFFERED BY BOZDOGAN.ASIMISMO, HAS BEEN ANALIZADO THE BEHAVIOR OF SELECTION CRITERIA FOR MODELS BASED ON THE KL-DIVERGENCIA (AIC, AICc And AICu) AND BASED ON THE J-DIVERGENCIA (KIC, KICc And KICu). THESE CRITERIA ARE NOT CONSISTENTES, COMPARATIVE CRITERIA HAVE BEEN BETWEEN THEMSELVES AND THEIR TIME WITH THE APPROACH Consistency BIC. THE STUDY HAS BEEN MADE IN THE CASE IN WHICH THE MODELS ARE CANDIDATES CONSTRUYEN AS MODELS ANIDADOS.

Author: FORS ALDRICH OCTAVI.
Year: 2005.
University: BARCELONA.
Place of defense: FACULTAT DE FÍSICA.
Place of preparation: FACULTAT DE FÍSICA.

Summary: In this work we have designed and developed a series of new techniques and tools for analyzing observational data into two distinct areas. On the one hand, the CCD image deconvolution great field (type survey). Moreover, the astrometry high resolution, and in particular the observational techniques of lunar occultations and interferometrías speckle. As to the first, have been implemented two deconvolution algorithms (maximum likelihood Richardson-Lucy (MLE) and its variant based adaptive wavelet (AWMLE)) to survey data from three projects: the Flagstaff Transit Telescope (FASTT), the Quasar Equatorial Survey Team (QUEST) and Near-Earth Space Surveillance Terrestrial (NESS-T). All three have been seen restrindia its size and resolution limits because of the way procurement drifft scanning (FASTT and QUEST), or very short relationship optical instrument NESS-T. It has been proposed and implemented a new methodology for the implementation of AWMLE and MLE paralas previous images. This makes it possible to assess the improvements made by the deconvolution in terms of increasing size and resolution limits. Moreover, it is quite general and is exportable to other survey data. The results show that AWMLE ensures an increase in the size limit of Am-0, 6mag and an improvement in the resolution limit of 1.0 pixel. Moreover, it appears that these trends are asintóticamente independent from a sufficiently large number of iterations. Simultaneously, it has been verified that the MLE deconvolution data FASTT not significantly affect the accuracy astrométrica nor introduced any bias toward the center pixel. As for the lunar occultations, has been designed, developed, implemented and evaluated a new technique for observing RCD lunar occultations. This based on the outline drifft scanning and allows show the intensity of the object concealed every 2ms. The technique allows virtually any observatory (professional or amateur) programs deal with lunar occultations purposes of scientific contribution. Simultaneously, it has conducted an observation lunar occultations (called CALOP) for 4 years and 71.5 nights has been executed in the Astronomical Observatory of Calar Alto, operating in the visible with CCD as in the ER with the camera MAGIC. As results of this program have been achieved stellar measuring 3 diameters on the order of 7mas and identify 15 new systems bianarios and a triple with angular separations of up 2mas. Finally, it has developed and implemented a new algorithm for automatic reduction of concealment based on wavelets. This algorithm has been successfully applied in reducing the overall occultations (400) registered in the program CALOP, and allows cope with the reduction of future campaigns observation mass. As for interferometry speckle has been designed, developed, implemented and evaluated a new observation technique for this type of CCD observations of high spatial resolution. It is based on the outline drift scanning and allows show the intensity of the object within the range of atmospheric coherence. It has been validated by measuring 4 binary systems orbiting known. The results of separation angle, angle and position of magnitude difference are in agreement with those reported by previous authors. The technique allows virtually any observatory (professional or amateur) face programs speckle for purposes of scientific contribution. Finally, it has been designed, implemented and validated a new technique autocalibración data speckle saving time observation point sources.

Author: OOLER PIQUE RAMON.
Year: 2005.
University: POLITÉCNICA DE CATALUÑA.
Place of defense: SALA D'ACTES DE LA FIME.
Place of preparation: OMEGA DESPATX 407, PLNATA 4º NORD.

Author: FERNÁNDEZ DEL POZO DE SALAMANCA JUAN ANTONIO.
Year: 2006.
University: POLITÉCNICA DE MADRID.
Place of defense: FACULTAD DE INFORMÁTICA.
Place of preparation: FACULTAD DE INFORMÁTICA.

Summary: The construction, operation and evaluation of Systems Decision Support through Networks Bayesianas and influence diagrams, among other models of reasoning, is managing tables with different information. These include tables of conditional probability, which specify the relationship between variables probabilistic, and the boards of optimal decisions resulting from evaluating the model. These tables, which can be highly complex, structured collected knowledge of the application domains on a set of variables probabilistic model figure. Under the name KBM2L introduce techniques to build the knowledge base of systems for decision support. We try to exploit the list KBM2L as a useful tool of knowledge representation system, the structure consists of graphical model, the model probability utility and the outcome of the evaluation. The graphical representation is qualitative and intuitive and therefore, system users have easy access to knowledge if they are experts in the problem. By contrast, quantitative models of probability and utility and evaluation show no knowledge easily. On the one hand, the models codify much of the knowledge numerically by numerous parameters, and secondly, tables optimal decisions are large. Both aspects prevent recognize the variables and relationships that describe the basic knowledge representation and explaining the results of the evaluation or inference. While the tables can be viewed as static entities, lists KBM2L are representations of knowledge dynamics. A specific configuration of the list determines the ability of explanation of knowledge, efficiency consultations to resolve the system of decision support from various angles, and the complexity of memory required to manage the knowledge base. The structure of the list to reveal the granularity of knowledge of the tables while the configurations we show the role that the variables of the model are in the evaluation inferred. The granularity enables articulate procedures to structure and better understand the knowledge that the system keeps in its tables. The role of the variables in different contexts and in the whole model provides us with a mechanism for generating explanations and knowledge of the proposals made by the system as well as sensitivity analysis of the model itself. The report is structured as follows. Following the introduction exposing the foundations of the list KBM2L as knowledge representation and describe the problems of finding the optimal representation and various proposed solutions. We are facing a combinatorial optimization problem we face in the ongoing metaheurísticas with algorithms and methods adapted to our goal. Here, we show the application of the techniques to optimal decisions tables and tables of conditional probability diagram of influence. Finally, we propose to conduct a sensitivity analysis of the model through natural extension of the list KBM2L usual with the parameters of interest.

Author: GUIJARRO MERELLES JOSÉ FERNANDO.
Year: 2006.
University: EXTREMADURA.
Place of defense: ESCUELA DE INGENIERÍAS INDUSTRIALES.
Place of preparation: ESCUELA DE INGENIERÍAS INDUSTRIALES DE LA UNIVERSIDAD DE EXTREMADURA.

Summary: This makes a doctoral thesis research work to prioritize the various factors that influence the certification processes of enterprises in accordance with ISO 9000, with particular focus on industrial companies. This first identified a total of seventeen factors involved in certification processes in accordance with the above standard, based on the more general models of quality management as proposed by the EFQM and the standards ISO 9000 in different versions, obtained by comparing the factors studied in the earlier investigations and adding some new ones as a result of recent changes to the rule. For the "hierarchy" of factors have been consulted directly to the certified companies, using as shown by industry in Extremadura. This has been achieved while conducting a detailed study of the status of certification under ISO 9000, among firms in Extremadura. An assessment of factors made for businesses has been achieved through a survey designed specifically for this job. With the data submitted for participants in the survey was achieved by adapting the Rasch model to this study, prioritize the importance of each of the factors studied. It used to the concept of propensity to certification, which measures the contribution of factors during the process of certification of enterprises, thus assigning a value to each of them. The methodology classified at the same time, each of the companies that have participated in the work for calculating each of them worth their propensity for certification. The conclusions of the thesis indicate the order of the factors to the extent obtained for each patterns and how they affect businesses differently, depending on the sector, the size, or type of activity.