Federal policies such as No Child Left Behind (NCLB) and Race to the Top (RTT) stand as examples of how teachers face increasing expectations that their activities be “data-based” or “data-driven.” Meeting these expectations requires assembling and analyzing a wide variety of data about students (e.g., demographics, discipline, locally designed tests, state test results, or longitudinal information). Computer data systems are commonly assumed to facilitate the work of educational data use. Indeed, the availability and computing power of these systems have continued to expand, further increasing the promises that these technologies hold for enhancing teaching and learning. Meaningful and widespread changes to teachers’ practices, however, have typically not occurred on a large-scale or systemic basis. Therefore, in this comparative case study of three school districts I examine the nature of districts’ efforts to improve teachers’ data use via computer data systems. I do so by examining the worldviews of various job roles in each district about data use and computer data systems. An erroneous assumption commonly made by districts was that these technologies are imbued with self-evident and predetermined effects on teacher work. Accordingly, the findings from this study speak to issues of sensemaking in districts. In them, I describe not only how teachers’ perspectives shaped their practices, but also how the alignment of perspectives among district roles influenced the implementation and success of district initiatives around computer data systems. As such, this study has implications for how districts plan, implement, and learn from initiatives around data use. / text

Background: Pelvic organ prolapse (POP) can be treated surgically or, more conservatively, with use of a pessary. Objective: To determine if the population of women treated for POP with the use of a pessary in one Nurse Practitioner's (NP) practice demonstrated health outcomes as better, same, or needing improvement through use of a data-based practice model from encounter data extracted from the electronic health record (EHR).Design: The project design was a quality improvement (QI) project, descriptive in nature. One Plan Do Study Act (PDSA) cycle was conducted for this QI project. Setting: NP managed specialty clinic in urban Southwestern Arizona that provides services to women with POP. Patients: Ten randomly selected women who had been treated conservatively for POP with use of a pessary were identified as two subpopulations and evaluated: women who received professional management of the pessary and women who were patient managed. Intervention: The intervention was the development of a data-based practice model, using patient profile data elements derived from the documented EHR encounters of the 10 women. Measurements: Twelve scales were developed to evaluate the patient profile data elements, generating numeric scores for each encounter. Two Decision Rules were then used to evaluate numeric scores by encounter, creating primary and secondary health outcomes. Limitations: Two limitations were identified. The QI project was limited by the small sample size of 10 patients. This is however, true to PDSA guidelines that recommend small scale cycles. The data were limited as only documented data were used. Conclusions: In general, the expected outcome was the outcome observed; the provider was unaware of any women in this QI Project who were not successfully treated with use of a pessary for treatment of POP. The value of this data-based practice model is that outcomes can be aggregated across populations rather than relying on recall of individual outcomes and therefore has potential to be used regularly and systematically as a quality feedback loop, as well as on a larger scale in future PDSA cycles to determine other outcomes beyond a single provider in this or other similar clinical populations.

No description available.

O projeto de controladores baseado em dados consiste no ajuste dos parâmetros do controlador diretamente das bateladas de dados do processo, sem a necessidade de um modelo. O ajuste é feito resolvendo um problema de otimização, onde procura-se o argumento que minimize uma determinada função custo. Para resolver o problema de otimização são utilizados nesses métodos o algoritmo do gradiente, o algoritmo de Newton e variações destes. O algoritmo do gradiente apenas necessita informação do gradiente da função custo enquanto que os outros utilizam mais informações como a hessiana. Para obter estas últimas informações são utilizados experimentos mais longos e mais complexos, o que torna a aplicação mais complicada. Nesta linha o algoritmo do gradiente se apresenta como a melhor alternativa, por este motivo foi escolhido como foco deste trabalho. A convergência do algoritmo do gradiente para o mínimo global da função custo, no contexto de projeto de controladores, não é encontrada na bibliografia, decidiu-se portanto estudá-la. Essa convergência depende das condições iniciais do algoritmo e do tamanho do passo de iteração utilizado. É mostrado que as condições iniciais precisam estar dentro de uma certa região de atração. Formas de aumentar esta região de atração são tratadas na metodologia chamada Shaping da Função Custo. A principal contribuição deste trabalho é apresentar um método eficiente para a escolha do tamanho do passo de iteração que garante a convergência para o mínimo global da função custo. Algumas informações do processo são necessárias para o cálculo do tamanho do passo de iteração, também são apresentadas maneiras de obter estimativas para estas informações. Simulações e experimentos demonstram o funcionamento dos métodos. / Data-based control design methods consist of adjusting the parameters of the controller directly from batches of input-output data of the process; no process model is used. The adjustment is done by solving an optimization problem, which searches the argument that minimizes a specific cost function. Iterative algorithms based on the gradient are applied to solve the optimization problem, like the steepest descent algorithm, Newton algorithm and some variations. The only information utilized for the steepest descent algorithm is the gradient of the cost function, while the others need more information like the hessian. Longer and more complex experiments are used to obtain more informations, that turns the application more complicated. For this reason, the steepest descent method was chosen to be studied in this work. The convergence of the steepest descent algorithm to the global minimum is not fully studied in the literature. This convergence depends on the initial conditions of the algorithm and on the step size. The initial conditions must be inside a specific domain of attraction, and how to enlarge this domain is treated by the methodology Cost Function Shaping. The main contribution of this work is a method to compute efficiently the step size, to ensure convergence to the global minimum. Some informations about the process are utilized, and this work presents how to estimate these informations. Simulations and experiments demonstrate how the methods work.

Controladores PID são amplamente utilizados no controle de processos industriais. Estes controladores precisam necessariamente ser ajustados adequadamente a fim de garantir a correta operação do processo controlado. A fim de suprir esta necessidade surgiram os chamados métodos de ajuste para controladores PID, inicialmente propostos por John Ziegler e Nathaniel B. Nichols em 1942. Desde então muitos outros métodos de ajuste baseados nas idéias de Ziegler e Nichols foram propostos, surgindo assim uma família de métodos afins. Em vista da simplicidade de implementação e do consequente baixo custo computacional envolvido, estes métodos mostraram-se adequados para serem incorporados ao firmware de controladores PID industriais de baixo custo. Estes métodos acabaram por gerar um legado tal que sua utilização persiste intensamente até os dias de hoje. No entanto, frente à crescente oferta de microcontroladores de baixo custo e alto desempenho, o custo computacional de um método de ajuste vem perdendo relevância. Isso abre margem para explorar outros métodos que proporcionem melhor desempenho e robustez, mas, por ventura, demandem mais recursos computacionais. Dessa forma, este trabalho propõe-se a avaliar métodos alternativos que sejam compatíveis com os recursos computacionais atuais. Métodos com maior custo computacional, como o Virtual Reference Feedback Tuning (VRFT), Iterative Feedback Tuning (IFT) e Iterative Correlation-based Tuning (ICbT), são apresentados como candidatos para serem incorporados ao firmware de controladores PID industriais. Tratam-se de métodos diretos de ajuste baseado em dados onde os parâmetros do controlador são determinados de forma que o comportamento do sistema em malha fechada seja tal que minimize um critério de desempenho definido a priori. Através deste critério de desempenho pode-se definir o comportamento desejado para o sistema em malha fechada. Neste trabalho são analisadas as principais características destes métodos, resultados obtidos e custo computacional. Com base nos resultados desta análise é mostrado que os métodos VRFT, IFT e ICbT podem ser utilizados como alternativa para o ajuste (incorporado ao firmware) de controladores PID industriais. / PID controllers are widely used in industrial process control. These controllers must necessarily be properly tuned to ensure the correct operation of the controlled process. In order to meet this need, the so-called tuning methods for PID controllers have emerged, initially proposed by John Ziegler and Nathaniel B. Nichols in 1942. Since then many other controller design methods based on the ideas of Ziegler and Nichols have been proposed, giving rise to a family of related methods. Given the simplicity of implementation and the low computational effort involved, these methods are suitable to be incorporated into the firmware of low cost industrial PID controllers. These methods have generated such a legacy that its use remains intense until the present day. However, with the growing offer of low cost and high performance microcontrollers, the computational effort of a tuning method is becoming less important. This opens up scope for exploring other methods that provide better performance and robustness, possibly at the cost of demanding more computational resources. This study aims to evaluate alternative methods that are compatible with current computational resources. Methods with higher computational effort, such as Virtual Reference Feedback Tuning (VRFT), Iterative Feedback Tuning (IFT) and Iterative Correlation based Tuning (ICbT) are presented as candidates to be incorporated into the firmware of industrial PID controllers. These are direct data-based methods for the adjustment of controllers where the parameters are determined such that the behavior of the closed-loop system is such as to minimize a performance criterion defined a priori. Through this performance criterion one can specify the desired behavior for the closed-loop system. This work analyzes the main characteristics of these methods, results and computational effort. Based on the results of this analysis it is shown that the methods VRFT, IFT and ICbT can be used as an alternative to the adjustment (build into the firmware) of industrial PID controllers.

xvi, 229 p. A print copy of this thesis is available through the UO Libraries. Search the library catalog for the location and call number. / This case study identified which of Baker and Richards' (2004) leadership models (compliance, performance, ecological) were used to make data-based decisions in six Oregon schools. Two elementary, two middle and two high schools in a suburban school district were selected. Typologies of each school's reported data Sources, Leadership, Processes and Impacts were developed. The results of the typologies were applied through pattern-matching to a Conceptual Model of Data-Based Schools developed by Hill (2004) in an earlier study. The study investigated (a) the similarities and differences in how the schools used the data they collected; (b) patterns that emerged indicating how data were used to inform decisions; and (c) the data-based leadership model (compliance, performance, ecological) evidenced at each school, school level and within the overall district. Findings indicated consistent patterns of data-based practices across all six schools and placed each of them, as well as the overall district, on the continuum between the performance and ecological leadership models. School administrators reported an ecological set of beliefs to guide their site-based decisions; teachers reported a performance set of beliefs and practices in their classrooms. There was no significant difference attributed to school levels. The findings build on Hill's (2004) previous study, strengthen Baker and Richards' (2004) ecological leadership model, and add to the emerging literature on ecological leadership in schools. School leaders can use the model to identify current practices in data-based decision-making and share their findings with their staff in order to improve data practices and move along the continuum toward ongoing and continuous school improvement. / Adviser: Diane Dunlap

The success of behavioral interventions depends not just on the quality of procedures employed but on the extent to which procedures are implemented. This study used a multiple-baseline across participants single-case design to assess the impact of an online data management application (the Individual Student Information System; ISIS- SWIS) on the fidelity and impact of individual student behavior support plans in typical school contexts. Three students with patterns of problem behavior and their supporting adults participated in the study. The research question examined if a functional relation exists between use of (a) performance self-assessment and (b) student impact assessment via ISIS-SWIS on the fidelity of behavior support plan implementation by adults and improvement in academic engagement and problem behavior by students. Results indicate the efficacy of ISIS-SWIS in improving treatment fidelity, decreasing student problem behavior, and increasing student academic engagement. Potential contributions of the study are discussed in terms of establishing efficient data systems for schools to use in monitoring staff and student behavior and using these data in a meaningful way that results in improved student outcomes and sustained behavior change.

O projeto de controladores baseado em dados consiste no ajuste dos parâmetros do controlador diretamente das bateladas de dados do processo, sem a necessidade de um modelo. O ajuste é feito resolvendo um problema de otimização, onde procura-se o argumento que minimize uma determinada função custo. Para resolver o problema de otimização são utilizados nesses métodos o algoritmo do gradiente, o algoritmo de Newton e variações destes. O algoritmo do gradiente apenas necessita informação do gradiente da função custo enquanto que os outros utilizam mais informações como a hessiana. Para obter estas últimas informações são utilizados experimentos mais longos e mais complexos, o que torna a aplicação mais complicada. Nesta linha o algoritmo do gradiente se apresenta como a melhor alternativa, por este motivo foi escolhido como foco deste trabalho. A convergência do algoritmo do gradiente para o mínimo global da função custo, no contexto de projeto de controladores, não é encontrada na bibliografia, decidiu-se portanto estudá-la. Essa convergência depende das condições iniciais do algoritmo e do tamanho do passo de iteração utilizado. É mostrado que as condições iniciais precisam estar dentro de uma certa região de atração. Formas de aumentar esta região de atração são tratadas na metodologia chamada Shaping da Função Custo. A principal contribuição deste trabalho é apresentar um método eficiente para a escolha do tamanho do passo de iteração que garante a convergência para o mínimo global da função custo. Algumas informações do processo são necessárias para o cálculo do tamanho do passo de iteração, também são apresentadas maneiras de obter estimativas para estas informações. Simulações e experimentos demonstram o funcionamento dos métodos. / Data-based control design methods consist of adjusting the parameters of the controller directly from batches of input-output data of the process; no process model is used. The adjustment is done by solving an optimization problem, which searches the argument that minimizes a specific cost function. Iterative algorithms based on the gradient are applied to solve the optimization problem, like the steepest descent algorithm, Newton algorithm and some variations. The only information utilized for the steepest descent algorithm is the gradient of the cost function, while the others need more information like the hessian. Longer and more complex experiments are used to obtain more informations, that turns the application more complicated. For this reason, the steepest descent method was chosen to be studied in this work. The convergence of the steepest descent algorithm to the global minimum is not fully studied in the literature. This convergence depends on the initial conditions of the algorithm and on the step size. The initial conditions must be inside a specific domain of attraction, and how to enlarge this domain is treated by the methodology Cost Function Shaping. The main contribution of this work is a method to compute efficiently the step size, to ensure convergence to the global minimum. Some informations about the process are utilized, and this work presents how to estimate these informations. Simulations and experiments demonstrate how the methods work.

Controladores PID são amplamente utilizados no controle de processos industriais. Estes controladores precisam necessariamente ser ajustados adequadamente a fim de garantir a correta operação do processo controlado. A fim de suprir esta necessidade surgiram os chamados métodos de ajuste para controladores PID, inicialmente propostos por John Ziegler e Nathaniel B. Nichols em 1942. Desde então muitos outros métodos de ajuste baseados nas idéias de Ziegler e Nichols foram propostos, surgindo assim uma família de métodos afins. Em vista da simplicidade de implementação e do consequente baixo custo computacional envolvido, estes métodos mostraram-se adequados para serem incorporados ao firmware de controladores PID industriais de baixo custo. Estes métodos acabaram por gerar um legado tal que sua utilização persiste intensamente até os dias de hoje. No entanto, frente à crescente oferta de microcontroladores de baixo custo e alto desempenho, o custo computacional de um método de ajuste vem perdendo relevância. Isso abre margem para explorar outros métodos que proporcionem melhor desempenho e robustez, mas, por ventura, demandem mais recursos computacionais. Dessa forma, este trabalho propõe-se a avaliar métodos alternativos que sejam compatíveis com os recursos computacionais atuais. Métodos com maior custo computacional, como o Virtual Reference Feedback Tuning (VRFT), Iterative Feedback Tuning (IFT) e Iterative Correlation-based Tuning (ICbT), são apresentados como candidatos para serem incorporados ao firmware de controladores PID industriais. Tratam-se de métodos diretos de ajuste baseado em dados onde os parâmetros do controlador são determinados de forma que o comportamento do sistema em malha fechada seja tal que minimize um critério de desempenho definido a priori. Através deste critério de desempenho pode-se definir o comportamento desejado para o sistema em malha fechada. Neste trabalho são analisadas as principais características destes métodos, resultados obtidos e custo computacional. Com base nos resultados desta análise é mostrado que os métodos VRFT, IFT e ICbT podem ser utilizados como alternativa para o ajuste (incorporado ao firmware) de controladores PID industriais. / PID controllers are widely used in industrial process control. These controllers must necessarily be properly tuned to ensure the correct operation of the controlled process. In order to meet this need, the so-called tuning methods for PID controllers have emerged, initially proposed by John Ziegler and Nathaniel B. Nichols in 1942. Since then many other controller design methods based on the ideas of Ziegler and Nichols have been proposed, giving rise to a family of related methods. Given the simplicity of implementation and the low computational effort involved, these methods are suitable to be incorporated into the firmware of low cost industrial PID controllers. These methods have generated such a legacy that its use remains intense until the present day. However, with the growing offer of low cost and high performance microcontrollers, the computational effort of a tuning method is becoming less important. This opens up scope for exploring other methods that provide better performance and robustness, possibly at the cost of demanding more computational resources. This study aims to evaluate alternative methods that are compatible with current computational resources. Methods with higher computational effort, such as Virtual Reference Feedback Tuning (VRFT), Iterative Feedback Tuning (IFT) and Iterative Correlation based Tuning (ICbT) are presented as candidates to be incorporated into the firmware of industrial PID controllers. These are direct data-based methods for the adjustment of controllers where the parameters are determined such that the behavior of the closed-loop system is such as to minimize a performance criterion defined a priori. Through this performance criterion one can specify the desired behavior for the closed-loop system. This work analyzes the main characteristics of these methods, results and computational effort. Based on the results of this analysis it is shown that the methods VRFT, IFT and ICbT can be used as an alternative to the adjustment (build into the firmware) of industrial PID controllers.

O projeto de controladores baseado em dados consiste no ajuste dos parâmetros do controlador diretamente das bateladas de dados do processo, sem a necessidade de um modelo. O ajuste é feito resolvendo um problema de otimização, onde procura-se o argumento que minimize uma determinada função custo. Para resolver o problema de otimização são utilizados nesses métodos o algoritmo do gradiente, o algoritmo de Newton e variações destes. O algoritmo do gradiente apenas necessita informação do gradiente da função custo enquanto que os outros utilizam mais informações como a hessiana. Para obter estas últimas informações são utilizados experimentos mais longos e mais complexos, o que torna a aplicação mais complicada. Nesta linha o algoritmo do gradiente se apresenta como a melhor alternativa, por este motivo foi escolhido como foco deste trabalho. A convergência do algoritmo do gradiente para o mínimo global da função custo, no contexto de projeto de controladores, não é encontrada na bibliografia, decidiu-se portanto estudá-la. Essa convergência depende das condições iniciais do algoritmo e do tamanho do passo de iteração utilizado. É mostrado que as condições iniciais precisam estar dentro de uma certa região de atração. Formas de aumentar esta região de atração são tratadas na metodologia chamada Shaping da Função Custo. A principal contribuição deste trabalho é apresentar um método eficiente para a escolha do tamanho do passo de iteração que garante a convergência para o mínimo global da função custo. Algumas informações do processo são necessárias para o cálculo do tamanho do passo de iteração, também são apresentadas maneiras de obter estimativas para estas informações. Simulações e experimentos demonstram o funcionamento dos métodos. / Data-based control design methods consist of adjusting the parameters of the controller directly from batches of input-output data of the process; no process model is used. The adjustment is done by solving an optimization problem, which searches the argument that minimizes a specific cost function. Iterative algorithms based on the gradient are applied to solve the optimization problem, like the steepest descent algorithm, Newton algorithm and some variations. The only information utilized for the steepest descent algorithm is the gradient of the cost function, while the others need more information like the hessian. Longer and more complex experiments are used to obtain more informations, that turns the application more complicated. For this reason, the steepest descent method was chosen to be studied in this work. The convergence of the steepest descent algorithm to the global minimum is not fully studied in the literature. This convergence depends on the initial conditions of the algorithm and on the step size. The initial conditions must be inside a specific domain of attraction, and how to enlarge this domain is treated by the methodology Cost Function Shaping. The main contribution of this work is a method to compute efficiently the step size, to ensure convergence to the global minimum. Some informations about the process are utilized, and this work presents how to estimate these informations. Simulations and experiments demonstrate how the methods work.