Zugl.: Konstanz, Univ., Diss., 2008

University, Diss., 2002--Dortmund.

University, Diss., 2003--Dortmund.

<p>Wind is used as an energy source all over the world. To be able to use this effectively, there is a need for as good forecasts and forecast models as possible. One of these models is Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS®) that is used to calculate short time forecasts. This model is used here to calculate wind speeds at two different areas in Västra Götaland, Bengtsfors and Vänersborg. There are also wind measurements with SODAR stations for these areas. The first part of this work is to investigate the difference between two model resolutions, 36 and 12 km, for the model results compared with the measurements. The comparison is done by calculating some different statistical values. The results of these parameters show that the difference between the two resolutions is fairly small and that the lower resolution gives a slightly better result.</p><p>The second and major part of this work is to use two different regression models to adjust the result of the forecast models to the result of the measurements. These regression models will then be possible to use even when there are no measurements to compare with. The idea of these regression models is to find a way to describe the difference between the result of the forecast model and the SODAR measurements. This difference is then subtracted from the result of the forecast model so that you get an adjustment and more accurate result. The first regression model calculates the difference according to time of the day, the other model calculates the difference according to the wind speed.</p><p>Furthermore, the measurements used are taken from 75 meters height above the ground. These are then compared to some different results from the forecast model, for example different model heights and different resolutions, and also the model results adjusted with the regression models. The comparison is done by calculating the same statistic values as before, both with and without an adjustment with the regression models, and also to look at histograms that show the distribution of the difference. It is shown that with the regression adjustment, there is a clear improvement of the statistical values compared to the original results of the forecasts. For example the value of the absolute mean difference is reduced with approximately 0.4-0.7 m/s with an adjustment of the regression model. The histograms clearly show that a more even distribution occurs after the adjustment with the regression models. From having a major part of the differences at 1-2 m/s to now having the major part at around 0 m/s and furthermore there is also generally a lower difference between the measurements and the results from the forecast model.</p>

Bachelor-Arbeit Univ. St. Gallen, 2005.

Bachelor-Arbeit Univ. St. Gallen, 2006.

University, Diss., 2003--Oldenburg. / Dateiformat: zip, Dateien in unterschiedlichen Formaten.

Support Vector Machines (SVMs) are rarely benchmarked against other classification or regression methods. We compare a popular SVM implementation (libsvm) to 16 classification methods and 9 regression methods-all accessible through the software R-by the means of standard performance measures (classification error and mean squared error) which are also analyzed by the means of bias-variance decompositions. SVMs showed mostly good performances both on classification and regression tasks, but other methods proved to be very competitive. / Series: Report Series SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

A general class of fluctuation tests for parameter instability in an M-estimation framework is suggested. The tests are based on partial sum processes of M-estimation scores for which functional central limit theorems are derived under the null hypothesis of parameter stability and local alternatives. Special emphasis is given to parameter instability in (generalized) linear regression models and it is shown that the introduced M-fluctuation tests contain a large number of parameter instability or structural change tests known from the statistics and econometrics literature. The usefulness of the procedures is illustrated using artificial data and data for the German M1 money demand, historical demographic time series from Großarl, Austria, and youth homicides in Boston. / Series: Report Series SFB "Adaptive Information Systems and Modelling in Economics and Management Science"

Zugl.: Hamburg, Helmut-Schmidt-Universiẗat, Diss.