DistributedDataMining/en: Unterschied zwischen den Versionen
Keine Bearbeitungszusammenfassung |
K (apps updated) |
||
| Zeile 28: | Zeile 28: | ||
|Country=Germany | |Country=Germany | ||
|Area=[[BOINC-Projects/en#Grid/Informatic|Grid/Informatic]] | |Area=[[BOINC-Projects/en#Grid/Informatic|Grid/Informatic]] | ||
|Windows=Time Series Analysis: Stock Price Prediction 4.30<br>Medical Data Analysis: Laryngeal video classification 1. | |Windows=Time Series Analysis: Stock Price Prediction 4.30<br>Medical Data Analysis: Laryngeal video classification 1.21 | ||
|Linux=Time Series Analysis: Stock Price Prediction 4.30<br>Medical Data Analysis: Laryngeal video classification 1. | |Linux=Time Series Analysis: Stock Price Prediction 4.30<br>Medical Data Analysis: Laryngeal video classification 1.22 | ||
|Mac= | |Mac= | ||
|64bit=Time Series Analysis: Stock Price Prediction 4.30 [linux]<br>Medical Data Analysis: Laryngeal video classification 1. | |64bit=Time Series Analysis: Stock Price Prediction 4.30 [linux]<br>Medical Data Analysis: Laryngeal video classification 1.22 [linux] | ||
|PS3= | |PS3= | ||
|ATI= | |ATI= | ||
Version vom 12. März 2011, 18:47 Uhr
 Time Series Prediction Stock Price Prediction (active) Part of our research is devoted to Time Series Analysis. Our focus is on forecasting economic time series such as DAX and Dow Jones. At first, we focused on the application of artificial neural networks to forecast time series. A detailed description on this approach, the design of the experimental setting as well as the results are presented in [5]. Later on, we applied support vector machines to avoid the high computational complexity of neural networks. The resulting forecasts are equally impressive even though the necessary computational costs can be decreased significantly. In 2008, we published two related studies [6] and [7]. We extended our studies by using various learning algorithms in order to determine there applicability for stock price prediction. After analyzing the obtained results we made two important observations: (i) the influence of the learning algorithm is much lower than expected, but instead (ii) the training window size has a stronger impact on the quality of the prediction. Since, so far, temporal effects are rarely addressed in the literature, we concentrate in our dDM-project on the study of these temporal aspects in time series analysis. Social Network Analysis Tanja Falkowski proposed DenGraph - a density-based graph clustering algorithm [1]. This algorithm is deployable for - among other things - Social Network Analysis. The following studies were part of her PhD theses that is published as a book. Temporal Dynamics of the Last.fm Music Platform (temporarily suspended) In this application we applied DenGraph-IO to detect and observe changes in the music listening behaviour of Last.fm users during a period of two years. The aim was to see, whether the proposed clustering technique detects meaningful communities and evolutions [2], [3]. Temporal Evolution of Communities in the Enron Email Data Set (finished) The collapse of Enron, a U.S. company honored in six consecutive years by "Fortune" as "America's Most Innovative Company", caused one of the biggest bankruptcy cases in US-history. To investigate the case, a data set of approximately 1.5 million e-mails sent or received by Enron employees was published by the Federal Energy Regulatory Commission. We've used the processing power of dDM to analyze the temporal evolution of communities extracted from these email correspondences [4]. |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
