Eilers, Dennis: Contributions to data analytics techniques with applications in forecasting, visualization and decision support. Gottfried Wilhelm Leibniz Universität, Diss., 2018, XVIII, 96 S. DOI: https://doi.org/10.15488/4006
Zusammenfassung: | |
This cumulative dissertation summarizes and critically discusses seven peer-reviewed publications where I was involved as a co-author. All publications contribute to data analytics techniques. The dissertation consists of four main sections. (1) Machine Leaning in Finance: In this section a Decision Support Algorithm based in Reinforcement Learning is introduced which filters rule-based trading decisions. We contribute to the literature by describing the implementation of the algorithm. We also provide empirical evidence of financial market anomalies. (2) Mining Customer Reviews: Opinions from customers about certain products are more and more expressed on social media platforms. Here we provide the first study which analyses YouTube comments as a data source for an aspect-based Sentiment Analysis. We also contribute to the literature by proposing a filtering method based on Google Trends which sorts product aspects according to their relevance for the customers. (3) Forecasting Resale Prices of Used Cars: In this section we show how to efficiently forecast resale prices of used cars with Artificial Neural Networks. We provide lessons learned about long-term forecasts. We also provide insights in the importance of certain independent factors which determine the resale price. (4) Visual Model Evaluation: The research in this section is mainly driven by the question of how to better incorporate human domain knowledge in data science. We develop a visualization technique based on heat maps which provides a more intuitive view on errors of a machine learning model. The visualization technique allows domain experts to discuss the results of machine learning models with data science experts on the same level of complexity. | |
Lizenzbestimmungen: | CC BY 3.0 DE |
Publikationstyp: | DoctoralThesis |
Publikationsstatus: | publishedVersion |
Erstveröffentlichung: | 2018 |
Die Publikation erscheint in Sammlung(en): | Wirtschaftswissenschaftliche Fakultät Dissertationen |
Pos. | Land | Downloads | ||
---|---|---|---|---|
Anzahl | Proz. | |||
1 | Germany | 459 | 37,23% | |
2 | United States | 226 | 18,33% | |
3 | Luxembourg | 75 | 6,08% | |
4 | No geo information available | 58 | 4,70% | |
5 | Russian Federation | 55 | 4,46% | |
6 | China | 46 | 3,73% | |
7 | United Kingdom | 29 | 2,35% | |
8 | Morocco | 21 | 1,70% | |
9 | Czech Republic | 21 | 1,70% | |
10 | France | 20 | 1,62% | |
andere | 223 | 18,09% |
Hinweis
Zur Erhebung der Downloadstatistiken kommen entsprechend dem „COUNTER Code of Practice for e-Resources“ international anerkannte Regeln und Normen zur Anwendung. COUNTER ist eine internationale Non-Profit-Organisation, in der Bibliotheksverbände, Datenbankanbieter und Verlage gemeinsam an Standards zur Erhebung, Speicherung und Verarbeitung von Nutzungsdaten elektronischer Ressourcen arbeiten, welche so Objektivität und Vergleichbarkeit gewährleisten sollen. Es werden hierbei ausschließlich Zugriffe auf die entsprechenden Volltexte ausgewertet, keine Aufrufe der Website an sich.