Refine
Document Type
- Doctoral thesis (2)
- Bachelor thesis (1)
Has Fulltext
- yes (3)
Is part of the Bibliography
- no (3)
Keywords
- Prognose (3) (remove)
Within the last decades, the number of social networks is growing fast. The competition of retaining the customers to grow their platform and increase their profitability is rising. That is why companies need to detect possible churners to retain these. The problem of predicting the users’ lifetime, churning users, and the reasons for churning can be tackled by using machine learning.
The goal of this bachelor thesis is to build machine learning models to predict user churn and the user lifetime within the social network Jodel, a location-based anonymous messaging application for Android and iOS.
To get the best possible prediction results, we have started with extensive literature research, whose approaches we have tested and added to a machine learning pipeline to build predictive models. With these models, we have investigated the performance after different observation time windows and have finally compared the strongest models to detect similarities and understand the insights to learn their behaviour.
The results of this thesis are machine learning models for a selected representative set of communities varying in size within the Kingdom of Saudi Arabia and a country model leveraging all data. These models are used for a regression task by predicting the lifetime of a user and a multi-label classification of a user into six different churn classes. Additionally, we have also given models for a binary classification, where the model will predict if the user will churn within a given time or not. These models have shown general strong predictive power, which is shrinking when limiting the observation time window. Especially the binary classification yielded high accuracy of over 99%.
The best models have been used for predicting user churn within other communities to detect communities with possible similar behaviour. These similarities then have been determined by features’ importance, where the most important features have got fed back into empirics. This has shown statistically significant differences between user groups with a different active time but as of today no clear trends were visible that had led us to define the communities’ behaviours.
Since the competition of social networks is still growing, the retaining of users will stay a core marketing strategy, which will need to be tackled by machine learning and artificial intelligence. The created models could be useful for predicting churning users within the platform Jodel to detect these customers that will churn within a given time.
Researches did not focus much on anonymous and location-based messaging. That is why the results of this thesis on the anonymous messaging application Jodel opens a variety of possible tasks for the future in this context.
Over the last decades, commodity futures markets have grown significantly because, in addition to hedgers using futures for risk management, investors have discovered the potential of futures in investment products. Motivated by this growing importance, this thesis is concerned with the analysis of risk (in terms of the established risk measure Expected Shortfall) and investment strategies in commodity futures markets.
First, we compare popular non-parametric estimators of Expected Shortfall (i. e., different variants of historical, outlier-adjusted and kernel methods) to each other, selected parametric benchmarks and estimates based on the idea of forecast combination within a multidimensional simulation setup (spanned by different distributional settings, sample sizes and confidence levels). We rank the estimators on the basis of classic error measures as well as an innovative performance profile technique, which we adapt from the mathematical programming literature. Our rich set of results supports academics and practitioners in the search for an answer to the question of which estimators are preferable under which circumstances.
After that, we present a full-scale analysis of Expected Shortfall in commodity futures markets. Besides illustrating the dynamics of historic Expected Shortfall, we evaluate whether popular estimators are suitable for forecasting future Expected Shortfall. By implementing a new backtest, we find that the performance of estimators hinges on market stability. Estimators tend to fail when markets are in turmoil and accurate forecasts are urgently needed. Even though a kernel method performs best on average, our results advise against the use of established estimators for risk (and margin) prediction.
Third, motivated by the deteriorating performance of traditional cross-sectional momentum strategies in commodity futures markets, we propose to resurrect momentum by incorporating autocorrelation information into the asset selection process. Put differently, we introduce measures of short and long memory (variance ratios and Hurst coefficients, respectively) telling us whether past winners and losers are likely to persist or not. Our empirical findings suggest that a memory-enhanced momentum strategy based on variance ratios significantly outperforms traditional momentum in terms of reward and risk, effectively prevents momentum crashes and is not bound to the movement of the overall commodity market. Furthermore, strategy returns cannot be explained by typical factor portfolios and macroeconomic variables and are robust to various parametrization choices, alternative data sets, transaction costs and data mining. Finally, and in contrast to a newly emerging strand of literature promoting the benefits of long memory measures in portfolio management, we show that Hurst coefficients do not carry investment-relevant information in a commodity momentum context.
Die vereinfachte Erfolgsformel des Managementansatzes Total Quality Management (TQM), oft auch Business Excellence genannt, lautet: Motivierte, zufriedene Mitarbeiter führen zu zufriedenen Kunden und damit zu hohen Gewinnen. Diese Erfolgsformel ist jedoch aufgrund hoher Komplexität der Wirkungen von TQM-Maßnahmen nicht für jedes Unternehmen gleichermaßen zutreffend. Gegenstand dieser Arbeit ist die Entwicklung einer Methode zur prognostischen Analyse unternehmensspezifischer Wirkungen von TQM-Maßnahmen. Diese Methode wurde auf Grundlage von Selbstbewertungen einiger Unternehmen im Rahmen von Qualitätspreisen in Deutschland entwickelt. Mit einer Sensitivitätsanalyse nach Vester wurde ein Wirkungsmodell für TQM-Maßnahmen erstellt und simuliert. Ein Moderationskonzept wurde entwickelt, um Simulation und Unternehmenskontext des jeweiligen Unternehmens miteinander zu verknüpfen. Ergebnis der Entwicklung ist eine dynamische Simulation, die es ermöglicht, die Komplexität von TQM und TQM-Maßnahmen in einem Unternehmen untersuchen zu können. Beabsichtigte und unbeabsichtigte Wirkungen von TQM-Maßnahmen auf verschiedene unternehmensinterne und -externe Bereiche können mit Hilfe der Methode vor der eigentlichen Umsetzung einer Maßnahme erkannt und die Maßnahmen entsprechend ihrer Zielsetzung angepasst werden.