Background: In a highly competitive world, leaders of firms highly dependent on innovation, such astelecom companies, must acquire data-driven managerial skills to systematically analyze datasets from multiple points of view to aid decision-making in the new context of Industry 4.0. Data mining can be performed on both tangible and intangible assets of Big Data sets, but systematic analytics performed on Small data can function as a crucial refinement for such insights. In addition, they are usable to train the algorithms during machine learning supervised stage, for example, when treating datasets in the field of psychometrics: originated by human perceptions and behaviors. This applies to the exploitation of strategic information, for business purposes, from intangible reservoirs, such as human capital aspects. Ambidexterity is a leadership conduct, primarily focusing on human capital and encompassing the behaviors of exploration and exploitation of new ideas. It has been historically proven to be essential for innovation. However, leaders and companies often limitedly focus on the exploitation of human capital aspects through psychometrics inserted in a data-driven framework. For business models that consider innovation as a matter to be pursued at any levels of the organization and not only confined to one specific department such as R&D, this is indeed a crucial element to be investigated to foster innovation and retaining a competitive edge. This research is performed in collaboration with a world leading telecom company and has been requested by its Innovation Leader. Objectives: The first objective of the research is to provide a flexible conceptual model and standardized methodology, suitable for incumbent, cross-country companies, highly dependent on innovation that intend to begin investigation on how those aspects influence their business performances. Second, the hypothesis testing of the conceptual model has the purpose of identifying the human capital aspects of national and corporate culture that show statistically significant andstronger cause-effect relationship towards enhancing innovation ambidexterity. Third, predictions interms of prevalence of explorative or exploitative innovative behaviors are aimed at providing indications on what the company could expect in terms of Innovation Ambidexterity with their current conditions. An automatable and replicable method that is data-driven-based for company`s decision makers is provided. It is also suitable for further integration within machine learning algorithms or simply as refinement of data mining insights and these aspects addressed are within the possibilities for improvement. The objective of the thesis is to test the methodology on a relatively small size sample to show to the company executives and Innovation Leader, the potential of the approach and the value that these data can have for decision making. They can decide to develop further the research involving larger samples at a later stage: inserting the analyses into an automatic periodical routine with dashboarding of the outcomes. During the post survey interviews, awareness among the management and executives has also been raised about the potential of such approach to obtain strategic business information unavailable until now. Please note that it was not the purpose of thisstudy to provide a conceptual model that was specific and suitable for the human capital`s characteristics of one specific company. The purpose was instead to provide a data-driven framework and a conceptual model that could be used by any company of the telecom sector to approach the task and to find moderating or mediating factors. It will also allow companies of different sectors to refine the model based on their needs at a later stage, as a possibility for future improvement. Methodology: A conceptual model, partially newly designed for this research is introduced. It incorporates selected elements of national and corporate culture appearing to be crucial for innovation ambidexterity, according to an extensive literature review. The quantitative analysis is also extensive.A less extensive analysis would have left too much uncertainty in the findings, undermining the confidence of executives in taking into consideration the results aimed at business actions. For these reasons, we recommend to researchers who are tackling the exploitation of intangible assets (such as human capital) to perform an extensive set of analyses. From the main dataset, the analyses of the methodology have been replicated on 5 sub-data sets based on the heterogeneity measured. The methodology includes CTA, PLS-SEM modeling on the outer model, PLS-SEM on the inner model including bootstrapping, MGA, FIMIX-PLS, IPMA, blindfolding for the predictive relevance of the model followed by POS and Weka predictions. Cause-effect relationships, mediating and moderating factors of national and internal culture have been also identified and indicated as part of the possiblefuture personalization of the model on the specific company`s human capital characteristics. The national culture attributes consist of power distance, uncertainty avoidance, collectivism, masculinity (unrelated to the gender) and gender diversity. The corporate culture attributes are categorized into caring climate, creative instability, boundary spanning, decision making and strategic horizon. The methodology employs a bottom-up survey design to collect data through an online questionnaire across three company sites located in Sweden, Italy, and China. The pieces of software used were SmartPLS 3 for Structural Equation Modeling and Predictions Oriented Segmentation and Weka 3.8.5 for a machine learning algorithm (an artificial neural network was used), as a double check on PLS-POS predictions. Some qualitative interpretations, pre and post survey interviews were also added. Results: Hypothesis testing and cross-comparisons are performed on groups such as employees, leaders, and the different geographical sites. During the evaluation of the results, special attention was put on the parameters related to the quality and statistical relevance, not only of the model tested on the six cohorts, but also on the single national and corporate attributes that build it up. The results show that explorative behaviors predict innovation ambidexterity to a larger extent than purely exploitative ones, confirming the main hypothesis. Predictions that were POS-based and verified by Weka machine learning algorithm have shown instead how the pursuit of innovation ambidexterity within the company is unbalanced towards exploitative behaviors. The study has provided PLS-SEM indications on how company executives may wish to pursue explorative behaviors towards innovation, but the company middle management is steering in the opposite direction, focusing on attributes more linked to efficiency and constant delivery. Consequently, what initially appeared to be a complex national culture issue of employees interfering with corporate culture, has been linked instead to a possible middle management issue related to two different business models: where one prevails over the other, instead of cooperating to reach innovation ambidexterity. This is a valuable strategic input for the company executives. The quantitative methodology uncovered results and patterns that the Innovation Leader had so far only intuitively perceived, and it offered such counterintuitive interpretation of the causes. With regards to national culture: power distance increases exploitative behaviors; gender diversity increases explorative behaviors, while it decreases exploitative behaviors. With regards to corporate culture: creative instability crucially increases explorative behaviors but decreases exploitative behaviors. Boundary spanning decreases exploitative behaviors. Conclusions: The thesis answered to the research question. It provided a scientific contribution, allowing a better understanding of how national and corporate cultures interact to generate explorativeand exploitative behaviors and ultimately innovation ambidexterity. It provided a flexible conceptual model and a standardized, automatable data-driven methodology suitable to discover insights from human capital aspects that influence innovation in a business: taking the analyses of human capital data performed by the firm “to the next level”. Recommendations for future research: A recommendation is to apply the proposed conceptual model to compare bigger size samples with even less heterogeneity, according to the optimal datasample`s characteristics identified. This will also allow a further personalization of the flexible andgeneral conceptual model presented (which is so far suitable for the general telecommunication sector), to more specific characteristics of the company which is the object of analysis. In a completely automated framework, it is also recommended to consider the possibilities of applying thisdata-driven, decision-making approach to other companies or industrial domains. This means, for example, integrating the proposed methodology within a machine learning algorithm in its supervised stage. The algorithm can be trained using the current analyses as refinement of insights provided by Big Data mining performed on sets related to innovation and collected within the firm`s organizational or production systems. It is also important to clarify that, according to the indication of the authors of this study, the results of the data-driven framework can be compared among different companies. However, to collect data from different companies through the same questionnaire shall be avoided because the quality of the results is highly dependent on the homogeneity of groups` mindsets and perceptions.

The intermittent nature of renewable energy, variations in energy demand, and fluctuations in oil and gas prices have all contributed to variable demand for power generation from coal-burning power plants. The varying demand leads to load-follow and on/off operations referred to as cycling. Cycling causes transients of properties such as pressure and temperature within various components of the steam generation system. The transients can cause increased damage because of fatigue and creep-fatigue interactions shortening the life of components. The data-driven model based on artificial neural networks (ANN) is developed for the first time to estimate properties of the steam generator components during cycling operations of a power plant. This approach utilizes data from the Coal Creek Station power plant located in North Dakota, USA collected over 10 years with a 1-hour resolution. Cycling characteristics of the plant are identified using a time-series of gross power. The ANN model estimates the component properties, for a given gross power profile and initial conditions, as they vary during cycling operations. As a representative example, the ANN estimates are presented for the superheater outlet pressure, reheater inlet temperature, and flue gas temperature at the air heater inlet. The changes in these variables as a function of the gross power over the time duration are compared with measurements to assess the predictive capability of the model. Mean square errors of 4.49E-04 for superheater outlet pressure, 1.62E-03 for reheater inlet temperature, and 4.14E-04 for flue gas temperature at the air heater inlet were observed.

Orientador: Paula Tavares Pinto / Resumo: A relação da Linguística de Corpus com o Ensino de Línguas, apesar de receber foco mesmo antes do advento dos computadores, se intensificou por volta da década de 90, momento em que pesquisas em corpora de aprendizes e em Aprendizagem Direcionada por Dados foram enfatizadas. Considerado esse estreitamento, esta pesquisa objetiva compilar quatro corpora de aprendizes a partir do uso autêntico da língua com o intuito de desenvolver atividades didáticas direcionadas por dados dos próprios alunos que promovam nos discentes um perfil autônomo de investigação linguística (mais precisamente das preposições with, in, on, at, for e to). No tocante à fundamentação teórica, destacam-se Prabhu (1987), Skehan (1996), Willis (1996), Nunan (2004) e Ellis (2006) a respeito do Ensino de Línguas por Tarefas, Jenkins (2012) e Neves (2014) que discorrem sobre as ficções de fã. Já sobre a Linguística de Corpus, tem-se Sinclair (1991), Berber Sardinha (2000) e Viana (2011). Granger (1998, 2002, 2013) mais relacionado a Corpus de Aprendizes, e Johns (1991, 1994), Berber Sardinha (2011) e Boulton (2010) no que diz respeito à Aprendizagem Direcionada por Dados. Como metodologia, levantaram-se textos escritos pelos alunos a partir de uma tarefa de produção escrita em que eles redigiram uma ficção de fã. Em seguida, esses textos formaram dois corpora de aprendizes iniciais, que foram analisados com o auxílio da ferramenta AntConc (ANTHONY, 2018) no intuito de observar a presença ou não de inadequações ... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: Although the relation between Corpus Linguistics and Language Teaching has been emphasized even before the advent of computers, it has been highlighted around the 90s. This was the moment when research on learner corpora and Data-Driven Learning was focused. Having said that, this study aimed to compile four learner corpora based on the authentic use of the language. This was done in order to develop data-driven teaching activities that could promote, among the students, an autonomous profile of linguistic investigation (more precisely about the prepositions with, in, on, at, for and to). Concerning the existing literature, we highlight the works of Prabhu (1987), Skehan (1996), Willis (1996), Nunan (2004) and Ellis (2006) about Task-Based Language Teaching, and Jenkins (2012) and Neves (2014) about fanfictions. In relation to Corpus Linguistics, this study is based on Sinclair (1991), Berber Sardinha (2000) and Viana (2011). Granger (1998, 2012, 2013) is referenced to define learner corpora, and Johns (1991, 1994), Berber Sardinha (2011) and Boulton (2010) to discuss Data-Driven Learning. The methodological approach involved the collection of the compositions from Language Teaching undergraduate students who developed a writing task in which they had to write a fanfiction. These texts composed two learner corpora, which were analyzed with the AntConc tool (ANTHONY, 2018) with the purpose of observing the occurrence of prepositions in English and whether they were accurately ... (Complete abstract click electronic access below) / Mestre

[pt] A série temporal é a representação mais comum para a evoluçãao no tempo de uma variável qualquer. Em um problema de previsão de séries temporais, procura-se ajustar um modelo para obter valores futuros da série, supondo que as informações necessárias para tal se encontram no próprio histórico da série. Como os fenômenos representados pelas séries temporais nem sempre existem de maneira isolada, pode-se enriquecer o modelo com os valores históricos de outras séries temporais relacionadas. A estrutura formada por diversas séries de mesmo intervalo e dimensão ocorrendo paralelamente é denominada série temporal multivariada. Esta dissertação propõe uma metodologia de geração de um Sistema de Inferência Fuzzy (SIF) para previsão de séries temporais multivariadas a partir de dados históricos, com o objetivo de obter bom desempenho tanto em termos de acurácia de previsão como no quesito interpretabilidade da base de regras – com o intuito de extrair conhecimento sobre o relacionamento entre as séries. Para tal, são abordados diversos aspectos relativos ao funcionamento e à construção de um SIF, levando em conta a sua complexidade e claridade semântica. O modelo é avaliado por meio de sua aplicação em séries temporais multivariadas da base completa da competição M3, comparandose a sua acurácia com as dos métodos participantes. Além disso, através de dois estudos de caso com dados reais públicos, suas possibilidades de extração de conhecimento são exploradas por meio de dois estudos de caso construídos a partir de dados reais. Os resultados confirmam a capacidade do AutoMFIS de modelar de maneira satisfatória séries temporais multivariadas e de extrair conhecimento da base de dados. / [en] A time series is the most commonly used representation for the evolution of a given variable over time. In a time series forecasting problem, a model aims at predicting the series future values, assuming that all information needed to do so is contained in the series past behavior. Since the phenomena described by the time series does not always exist in isolation, it is possible to enhance the model with historical data from other related time series. The structure formed by several different time series occurring in parallel, each featuring the same interval and dimension, is called a multivariate time series. This dissertation proposes a methodology for the generation of a Fuzzy Inference System (FIS) for multivariate time series forecasting from historical data, aiming at good performance in both forecasting accuracy and rule base interpretability – in order to extract knowledge about the relationship between the modeled time series. Several aspects related to the operation and construction of such a FIS are investigated regarding complexity and semantic clarity. The model is evaluated by applying it to multivariate time series obtained from the complete M3 competition database and by comparing it to other methods in terms of accuracy. In addition knowledge extraction possibilities are explored through two case studies built from actual data. Results confirm that AutoMFIS is indeed capable of modeling time series behaviors in a satisfactory way and of extractig meaningful knowldege from the databases.

Rapid digital transformation, accelerated by covid-19 and a younger, more digital workforce, has changed the B2B sales environment affecting customer behavior, business practices and technologies. To adapt to this change B2B companies need to create new endeavors that connect marketing and sales activities, and identify how these can be efficiently enhanced by technology and digital tools. There is a need to identify effective tactics in different phases of the marketing and sales process. Consequently, more academic research is needed to investigate critical issues that technology may have in the B2B buying process. This paper thereby aims to establish a framework for B2B companies on how to optimize the customer journey by supporting the sales process with digital inbound marketing. This is examined by looking at how B2B companies can optimize the sales process by targeting customers’ needs with relevant actions throughout the customer journey with the help of technology and digital means. The results of this exploratory single case study demonstrates that the customer journey is complex with several touchpoints, and to create optimized processes, the customer journey must be adapted to each specific customer segment. Further, the study also contributes to the literature by demonstrating the importance of the marketing and sales departments being integrated and working together to create efficient processes. The presented framework for a customer journey can be used by managers to visualize the sales process, and identify at what stages the process can be made more efficient by digital means.

Decisions have been automated since the early days. Ever since the rise of AI, ML and DataAnalytics, algorithmic decision-making has experienced a boom time. Nowadays, using AI withina company is said to be critical to the success of a company. Considering the point that it can bequite costly to develop AI/ ML and integrating it into decision-making, it is striking how littleresearch was put into the identification and analysis of its cost drivers by now. This thesis is acontribution to raise and the awareness of possible cost drivers to algorithmic decisions. Thetopic was divided in two subgroups. That is solely algorithms and hybrid decision-making. Asystematic literature review was conducted to create a theoretical base for further research. Thecost drivers for algorithms to make decisions without human interaction, the identified costdrivers identified can be found at Data Storage (including initial, floor rent, energy, service,disposal, and environmental costs), Data Processing, Transferring and Migrating. Additionally,social costs and the ones related to fairness as well as the ones related to algorithms themselves(Implementation and Design, Execution and Maintenance) could be found. Business Intelligenceused for decision making raises costs in Data quality, Update delays of cloud systems, Personneland Personnel training, Hardware, Software, Maintenance and Data Storage. Moreover, it isimportant to say that the recurrence of some costs was detected. Further research should go inthe direction of applicability of the theoretical costs in practice.

Large and complex software that is frequently changed leads to testing challenges. It is well established that the later a fault is detected in software development, the more it costs to fix. This thesis aims to research and develop a method of generating relevant and non-redundant test cases for a regression test suite, to catch bugs as early in the development process as possible. The research was executed at Axis Communications AB with their products and systems in mind. The approach utilizes user data to dynamically generate a Markov chain model and with a Markov chain Monte Carlo method, strengthen that model. The model generates test case proposals, detects test gaps, and identifies redundant test cases based on the user data and data from a test suite. The sampling in the Markov chain Monte Carlo method can be modified to bias the model for test coverage or relevancy. The model is generated generically and can therefore be implemented in other API-driven systems. The model was designed with scalability in mind and further implementations can be made to increase the complexity and further specialize the model for individual needs.

Reconstruction of sonar images is an inverse problem, which is normally solved with model-based methods. These methods may introduce undesired artifacts called angular and range leakage into the reconstruction. In this thesis, a method called Learned Primal-Dual Reconstruction, which combines a data-driven and a model-based approach, is used to investigate the use of data-driven methods for reconstruction within sonar imaging. The method uses primal and dual variables inspired by classical optimization methods where parts are replaced by convolutional neural networks to iteratively find a solution to the reconstruction problem. The network is trained and validated with synthetic data on eight models with different architectures and training parameters. The models are evaluated on measurement data and the results are compared with those from a purely model-based method. Reconstructions performed on synthetic data, where a ground truth image is available, show that it is possible to achieve reconstructions with the data-driven method that have less leakage than reconstructions from the model-based method. For reconstructions performed on measurement data where no ground truth is available, some variants of the learned model achieve a good result with less leakage.

The purpose of this study is to produce new knowledge concerning the perceived user’s value generated using machine learning technologies that activate data network effects factors that create value through various business model themes. The data network effects theory represents a set of factors that increase the user’s perceived value for a platform that uses artificial intelligence capabilities. The study followed an abductive research approach where initially found facts were matched against the data network effects theory to be put in context and understood. The study’s data was gathered through semi-structured interviews with experts who were active within the research area and chosen based on their practical experience and their role in the digitization of the healthcare sector. The results show that three out of six factors were fully realized contributing to value creation while two of the factors showed to be partially realized in order to contribute to value creation and that is justified by the exclusion of users' perspectives in the scope of the research. Lastly, only one factor has limited contribution to the value creation due to the heavy regulations limiting its realization in the health sector. It is concluded that data network effects moderators contributed differently in the activation of various business model themes for value creation in a general manner where further studies should apply the theory in the assessment of one specific AI health offering to take full advantage of the theory potential. The theoretical implications showed that the data network factors may not necessarily be equally activated to contribute to value creation which was not initially highlighted by the theory. Additionally, the practical implications of the study’s results may help managers in their decision-making process on which factors to be activated for which business model theme.

Debris-slides are fast-moving landslides that occur in the Appalachian region including the Great Smoky Mountains National Park (GRSM). Various knowledge and data-driven approaches using spatial distribution of the past slides and associated factors could be used to estimate the region’s debris-slide susceptibility. This study developed two debris-slide susceptibility models for GRSM using knowledge-driven and data-driven methods in GIS. Six debris-slide causing factors (slope curvature, elevation, soil texture, land cover, annual rainfall, and bedrock discontinuity), and 256 known debris-slide locations were used in the analysis. Knowledge-driven weighted overlay and data-driven bivariate frequency ratio analyses were performed. Both models are helpful; however, each come with a set of advantages and disadvantages regarding degree of complexity, time-dependency, and experience of the analyst. The susceptibility maps are useful to the planners, developers, and engineers for maintaining the park’s infrastructures and delineating zones for further detailed geotechnical investigation.