Search

This study aims to analyze the gap between the level of heat wave adaptation policies and heat wave effects in South Korean municipalities. First, the types of industries in municipalities were classified using factor analysis and cluster analysis. Second, the level of heat wave adaptation policy in the municipalities was assessed using a fuzzy analytic hierarchy process analysis. Third, the gap between the level of heat wave adaptation policy and the heat wave effect was analyzed. The results show that the heat wave adaptation policies were established in accordance with the heat wave effects to at least some degree. However, closer to the long-term future (2095), the policies have not sufficiently matched the level of heat wave effects. The proportion of municipalities with insufficient levels of heat wave adaptation policies against the heat wave effects was higher among urban-type municipalities. The analysis results suggest two policy implications. First, the heat wave adaptation policies of municipalities should be established through continuous feedback on the predictions of future heat wave effects. Second, urban-type municipalities should strengthen their planning authority and competence by securing a professional workforce and budgets for the establishment of heat wave adaptation policies.

In mathematical optimisation there is a rich interaction between theory and application. Theory provides tools to solve real life problems, which in return motivate further theoretical research. This doctoral thesis presents results of the different aspects of mathematical optimisation. In the first part we investigate the dissociation set problem, a theoretical problem on graphs. We prove several bounds and show structural properties. In the next part we consider clusterings. We provide approximation guarantees for the clustering volume and analyse the consequences of hierarchical clusterings. In the third part we apply mathematical optimisation to improve the quality of micro X-ray fluorescence measurements. Finally, in the last part we consider two real life optimisation problems that occur in the production of a medium-sized company.

In recent years, early retirement decisions have become more frequent in the European Union despite political efforts to prevent early retirement. This is a growing problem for the social security system. The study focuses on a life course approach using employment biographies and investigates the influence of work stress and health on early retirement intentions. Data of employees who were born in either 1959 or 1965 of the German cohort study on work, age, health and work participation are analysed (n = 3338). By linking survey and register data from 1993 to 2011, a sequence analysis is conducted to identify employment biographies. To analyse the relationship between the employment biographies and intended early retirement, a longitudinal path analysis is computed and includes work stress, measured through effort-reward imbalance, and self-rated health. The statistical analyses identify three adverse employment biographies, i.e. part-time work, episodes of unemployment or marginal employment. In addition, two favourable employment biographies are determined, characterised by full-time work and few episodes of unemployment. The results of the path analysis show that employment biographies with high work-related stress have early retirement intentions. Among adverse employment biographies, indirect effects of poor health on the association between work stress and early retirement intentions are found. Unexpectedly, among full-time workers, work stress is also associated with early retirement intentions with an additional mediation through health. The findings of this study highlight the importance of the life course perspective when analysing retirement decisions. In addition to health-promoting interventions in the labour market, effects of psychosocial factors should be focussed on in order to reduce early exits from the labour market., publishedVersion

Machine learning and deep learning present a set of powerful modeling approaches, that in recent years have shown significant improvements across different fields of application. In the discipline of actuarial science, so far comparably little work has been done to explore these methods across the variety of different use cases. The present, cumulative thesis contributes to the field of applied machine and deep learning, as well as actuarial science. The research papers included in this work each address one specific, actuarial application, namely the grouping of insurance contracts, the modeling and analysis of surrender in life insurance and the calibration of hidden, inhomogeneous Markov chains. In each paper, building upon a theoretical framework, we develop application-specific models which are then tested for practical data. Overall, this thesis provides valuable insight on the applicability and benefit of machine and deep learning for distinct actuarial challenges and points out promising directions for future research.

Gegenstand dieser Arbeit sind Algorithmen für zwei fundamentale Analysemethoden von sequentiellen Daten räumlicher Natur. Diese Art von Daten entsteht zum Beispiel, wenn Beobachtungen von räumlicher Natur wiederholt gemessen werden, so wie ein GPS Sensor die Position eines sich bewegenden Objekts misst. Ein anderes Beispiel sind sequentielle Daten, die üblicherweise ihrer Form nach analysiert werden, wie Temperatur-, Druck- oder Spannungskurven. Die erste Analysemethode ist die Zusammenfassung einer gegebenen Menge von Folgen von Punkten aus einem metrischen Raum durch eine einzelne Punktfolge, die die gemeinsamen Charakteristiken aggregiert. Die zweite betrachtete Analysemethode ist das sogenannte Clustering. Hier ist eine Menge von Punktfolgen gegeben und es soll eine aussagekräftige Partitionierung berechnet werden. Genauer soll die gegebene Menge in Cluster aufgespalten werden, sodass sich die Elemente innerhalb eines Clusters eine gemeinsame Aggregatfolge teilen.

International audience; Recent literature recognises the importance of situating social networks in spatial context. Yet, the spatial analysis of personal networks has often been limited to examining residential distances between actors. While distance is a central characteristic of social relationships, it is a poor indicator for understanding the intricacies of the geographical space, places and personal networks. This study develops an original approach for mapping and analysing personal networks based on their geographical scope and the distribution of the residential locations of network members in relevant geographical areas. We perform a factor and cluster analysis to identify the major geographical patterns of personal networks using two samples of egocentric networks from France and Switzerland. We validate the approach first by interpreting the patterns both quantitatively and qualitatively, and second by examining how these patterns relate to important social characteristics of respondents and their personal networks. We conclude by discussing the significance of this approach for integrating geographical information into the analysis of personal networks and for rethinking networks and the geographical space as co‐constituted.

With respect to the fulfillment of the objectives of the Europe 2020 strategy, the threat of poverty and social exclusion has not been sufficiently reduced in the European Union (EU) over the past decade, and large regional disparities persist. Young people are the most affected by the problems of income poverty, material deprivation and labour market exclusion, which are the three dimensions of poverty and social exclusion. In this article, we focus on comparing the EU countries in terms of the three listed dimensions, while revealing similarities and differences in the incidence and severity of these social phenomena among youth. In addition to measuring dimensions by the currently used AROPE (at risk of poverty or social exclusion) rate, we also use a larger spectrum of relevant indicators for a more comprehensive analysis. While the AROPE aggregate indicator uses the same methodology for the population of young people as for the whole population, our approach includes indicators that are specific to young people. We assume that all dimensions affect each other, so we apply multidimensional statistical methods such as principal components and cluster analysis to analyse them. These methods have revealed that some dimensions affect poverty and social exclusion to a greater extent and others to a lesser extent than might appear to be the case, based on AROPE’s partial rates. Moreover, we present quantified integral indicators that together with the results of the multivariate methods, provide a rather complex picture concerning the geographical distribution of poverty and social exclusion, as well as their dimensions in the EU, for the population of persons aged 18–24 years in 2008 and 2017.

Die gemischt-ganzzahlige lineare Optimierung ist im Stand der Forschung für Planungsprobleme technischer Systeme erprobt, benötigt jedoch für komplexere Systeme zu hohe Rechenzeiten. Deshalb wird im Rahmen der Dissertation ein Framework zur Bildung und Bewertung von Typperioden im Optimierungskontext entwickelt, um die Rechenzeiten systematisch zu reduzieren. Dabei liegt der Fokus auf der Auswahl und Entwicklung geeigneter Methoden zur Datenvorbereitung, Normierung, Zeitreihentransformation, Clustering-Algorithmen-Auswahl und Bewertung. Das Framework wird an einem neuartigen Optimierungsmodell zur Auslegung der gekoppelten Produktion von Stahl und Methanol mit einem am deutschen Intraday-Spotmarkt betriebenen Elektrolyseur exemplarisch angewendet. Im Vergleich zu etablierten, mathematischen Bewertungsmethoden erweisen sich die entwickelten Dauerlinien als besonders geeignet, um die Genauigkeit der resultierenden Optimierungsergebnisse zu prognostizieren.

Die im Forschungsfeld der industriellen Beziehungen häufige Methodik der Fallstudie weist neben Vorteilen mit der oft ungeklärten Verortung im großzahligen Bild der Statistik einen gravierenden Nachteil auf. Im Kontext von Mixed Method-Designs wird die Möglichkeit der gezielten Auswahl von Fällen auf der Basis von Clusteranalysen erörtert, um typische Fallstudien zu gewinnen. Wenn sich die sequentielle Vorgehensweise als zu zeitintensiv erweist, können simultane Erhebungen oder auch großzahlige Sekundärdaten und Fallstudien sowie überlappende Informationen zur Verortung von Fällen in typologischen Analysen genutzt werden. The case study methodology, which is frequently used in the research field of industrial relations, has advantages. A serious disadvantage in that it is often unclear where it is located in the large-scale statistical picture. In the context of mixed method designs, the possibility of the purposive selection of cases on the basis of cluster analyses is discussed in order to obtain typical case studies. If the sequential approach proves to be too time-consuming, simultaneous surveys or large-scale secondary data and case studies as well as overlapping information can be used to locate cases in typological analyses.

Modeling text data is becoming increasingly popular. Topic models and in particular the latent Dirichlet allocation (LDA) represent a large field in text data analysis. In this context, the problem exists that running LDA repeatedly on the same data yields different results. This lack of reliability can be improved by repeated modeling and a reasonable choice of a representative. Further, updating existing LDA models with new data is another common challenge. Many dynamic models, when adding new data, also update parameters of past time points, thus do not ensure the temporal consistency of the results. In this cumulative dissertation, I summarize in particular my methodological papers from the two areas of improving the reliability of LDA results and updating LDA results in a temporally consistent manner for use in monitoring scenarios. For this purpose, I first introduce the state of research for each of the two areas. After explaining the idea of the corresponding method, I give examples of applications in which the method has already been used and explain the implementation as an R package. Finally, for both fields I provide an outlook on potential further research., Die Modellierung von Textdaten erfährt wachsende Popularität. Einen großen Bereich in der Textdatenanalyse bilden Topic Modelle und dabei im Speziellen das Modell latent Dirichlet allocation (LDA). Dabei existiert die Problematik, dass sich bei einer wiederholten Ausführung der LDA auf denselben Daten verschiedene Resultate ergeben. Dieser Mangel an Reliabilität lässt sich durch eine wiederholte Modellierung und eine sinnvolle Wahl eines Repräsentanten verbessern. Eine weitere Herausforderung stellt das Aktualisieren von bestehenden LDA-Modellen anhand neuer Daten dar. Viele dynamische Modelle aktu- alisieren im Falle einer Hinzunahme neuer Daten auch Parameter vergangener Zeitpunkte und verletzen damit die zeitliche Konsistenz der Ergebnisse. In dieser kumulativen Dissertation fasse ich insbesondere meine methodischen Paper aus den beiden Themenbereichen der Verbesserung der Reliabilität von LDA-Ergebnissen und der zeitlich konsistenten Aktualisierung von LDA-Ergebnissen zur Nutzung in Monitoring- Szenarien zusammen. Dafür stelle ich zunächst jeweils den Forschungsstand dar. Nach einer Erläuterung der Idee der Methode, werden jeweils Beispiele gegeben, in denen die Methode bereits Anwendung fand und die Implementierung als R Paket erläutert. Zuletzt gebe ich für beide Themenbereiche einen Ausblick auf mögliche weitere Forschung.

Currently, Digital Twins receive considerable attention from practitioners and in research. A Digital Twin describes a concept that connects physical and virtual objects through a data linkage. However, Digital Twins are highly dependent on their individual use case, which leads to a plethora of Digital Twin configurations. Based on a thorough literature analysis and two interview series with experts from various electrical and mechanical engineering companies, this paper proposes a set of archetypes of Digital Twins for individual use cases. It delimits the Digital Twins from related concepts, e.g., Digital Threads. The paper delivers profound insights into the domain of Digital Twins and, thus, helps the reader to identify the different archetypical patterns., Business & information systems engineering;Bd 64. 2022, 3, S. 375-391

This thesis is concerned with two classes of optimization problems which stem mainly from statistics: clustering problems and cardinality-constrained optimization problems. We are particularly interested in the development of computational techniques to exactly or heuristically solve instances of these two classes of optimization problems. The minimum sum-of-squares clustering (MSSC) problem is widely used to find clusters within a set of data points. The problem is also known as the $k$-means problem, since the most prominent heuristic to compute a feasible point of this optimization problem is the $k$-means method. In many modern applications, however, the clustering suffers from uncertain input data due to, e.g., unstructured measurement errors. The reason for this is that the clustering result then represents a clustering of the erroneous measurements instead of retrieving the true underlying clustering structure. We address this issue by applying robust optimization techniques: we derive the strictly and $\Gamma$-robust counterparts of the MSSC problem, which are as challenging to solve as the original model. Moreover, we develop alternating direction methods to quickly compute feasible points of good quality. Our experiments reveal that the more conservative strictly robust model consistently provides better clustering solutions than the nominal and the less conservative $\Gamma$-robust models. In the context of clustering problems, however, using only a heuristic solution comes with severe disadvantages regarding the interpretation of the clustering. This motivates us to study globally optimal algorithms for the MSSC problem. We note that although some algorithms have already been proposed for this problem, it is still far from being “practically solved”. Therefore, we propose mixed-integer programming techniques, which are mainly based on geometric ideas and which can be incorporated in a branch-and-cut based algorithm tailored to the MSSC problem. Our numerical experiments show that these techniques significantly improve the solution process of a state-of-the-art MINLP solver when applied to the problem. We then turn to the study of cardinality-constrained optimization problems. We consider two famous problem instances of this class: sparse portfolio optimization and sparse regression problems. In many modern applications, it is common to consider problems with thousands of variables. Therefore, globally optimal algorithms are not always computationally viable and the study of sophisticated heuristics is very desirable. Since these problems have a discrete-continuous structure, decomposition methods are particularly well suited. We then apply a penalty alternating direction method that explores this structure and provides very good feasible points in a reasonable amount of time. Our computational study shows that our methods are competitive to state-of-the-art solvers and heuristics.

Purpose: The aim of this paper is to analyse comparatively, at regional level, the current state of a wide range of indicators of Vocational Education and Training (VET) at regional level in Spain. This will make it possible to characterise and better understand the existence of a variety of regional VET systems, including the Initial VET and Continuous VET subsystems within Spain, doing so under a multidimensional approach examining VET supply and demand. Methods: Systemic analysis of Spanish VET indicators leads to a selection of 54 indicators, which are then compared at regional level using k-means clustering. This approach identifies similarities and differences (clusters) across all of Spain's 17 autonomous communities (Spanish regions). The correlation between the variables is then analysed to examine the interaction between the VET system's supply and demand dimensions. Findings: The results show that 19 indicators explain the main differences between autonomous communities, which form two distinct clusters. Both VET supply and VET environment and demand influence cluster formation and inter-cluster differences. In the set of indicators that differentiate the two clusters of autonomous communities, close interaction is detected between certain indicators of VET supply and demand, especially those referring to the STEM occupational group, which confirms the correlation between these dimensions, albeit to a limited extent. Conclusions: While it is necessary to analyse the differences between clusters in greater depth, the results suggest that Spain’s regions are split into two distinct groups (clusters) in which the respective VET systems are developing and advancing differently. Moreover, there is evidence of a small number of significant interrelations between indicators of VET supply and demand, which point to both the VET system's specialisation and its inclusive nature.

Die städtische Verkehrsplanung und Verkehrspolitik wird häufig anhand von Städtevergleichen des Verkehrsverhaltens ihrer Einwohner bewertet. Eine Städtetypologie kann dazu beitragen, diese Unterschiede zutreffend zu interpretieren. Dazu werden in diesem Beitrag für 44 deutsche Städte 27 Indikatoren der sozioökonomischen und räumlichen Strukturen, des Verkehrsverhaltens und der Bewertung der Verkehrssituation genutzt. Aus einer Faktoren- und Clusteranalyse werden mehrere Städtecluster abgeleitet. Als latente Variablen erweisen sich die Faktoren metropolitane Prägung, Fahrradklima, Wohlstand, studentische Prägung und die Orientierung am Privat-Pkw. Die anschließende Clusteranalyse führt zu fünf Typen. In dieser Typologie gehören die meisten Städte, die in Deutschland als gute Beispiele kommunaler Verkehrspolitik gelten, zum Cluster der Metropolen oder zum Cluster der Universitätsstädte mit positivem Fahrradklima., Comparisons of cities are frequently used to evaluate urban transportation planning and policies. A typology of cities can contribute to adequately interpret the identified differences between cities. For this purpose, this paper collects 27 indicators of socio-economic characteristics, spatial structures, travel behaviour and the evaluation of local transport situation for 44 major German cities. We generate clusters of cities based on a factor and cluster analysis. The factor analysis identifies five latent variables called metropolitan character, bicycle-friendliness, affluence, influence of students, and private car orientation. The subsequent cluster analysis identifies five clusters. According to this typology, most cities that are considered good practices of urban transport policy in Germany belong either to the cluster of metropolitan cities or university cities with a cycling-friendly environment., Raumforschung und Raumordnung = Spatial research and planning;80(2)

Urban areas are major consumers of environmental resources and thus often place unsustainable demands on natural resources. As half of the world’s population (55%) lives in urban areas, the environmental degradation produced by cities threatens the health and quality of life of a fair share of the world’s population. For these reasons, progress towards sustainable urban development must be monitored and measured through suitable indicators. With reference to the assessment of air quality as a specific dimension of environmental quality in urban areas, existing studies have introduced various methodologies that mostly focus on objective measures (typically, exposure to outdoor air pollutants) while neglecting measures based on individual perceptions. Our goal is to contribute to filling this gap. To this end, we explore the relationship between objective and subjective measures of urban air quality in European countries. While the objective indicator is based on concentrations of PM2.5, our subjective indicator is reconstructed from individual perceptions collected through the European Union Statistics on Income and Living Conditions (EU-SILC) sample survey. Finally, through a cluster analysis, we classify the countries into homogeneous groups based on the values of these indicators. Our analysis reveals several differences in the country rankings according to the two indicators. For one group of countries, both approaches converge, thus leading to more definitive conclusions. For other countries, the mismatch between the two indicators suggests that either approach alone is not able to capture the full picture on air quality in urban environments.

Forschungsleistungen sind nicht unabhängig von den gegebenen institutionellen Rahmenbedingungen. Auf Basis der personellen und finanziellen Ausstattung sowie der Lehrbelastung identifizieren wir sechs Gruppen der universitären Wirtschaftswissenschaften in Deutschland. Für diese vergleichen wir Promotionen, Drittmittel, Publikationen und die Platzierung im RePEc-Ranking als gängige Indikatoren zur Forschungsleistung. Die Ergebnisse bestätigen erwartungsgemäß Zusammenhänge zwischen Input- und Outputgrößen. Besonders bedeutsam ist die Ausstattung der Professuren mit wissenschaftlichem Personal. Zugleich lassen sich gemeinsame Forschungsprofile in den verschiedenen Gruppen erkennen: z.B. sind einige wenige (tendenziell große und junge) Standorte auf Top-Publikationen spezialisiert; andere sind darauf weniger ausgerichtet, leisten aber vielzitierte und drittmittelstarke Forschung. Die Analyse unterstreicht die notwendige Differenzierung bei Leistungsvergleichen und soll als Diskussionsanstoß für Weiterentwicklungen dienen.

Complex interactive test items are becoming more widely used in assessments. Being computer-administered, assessments using interactive items allow logging time-stamped action sequences. These sequences pose a rich source of information that may facilitate investigating how examinees approach an item and arrive at their given response. There is a rich body of research leveraging action sequence data for investigating examinees' behavior. However, the associated timing data have been considered mainly on the item-level, if at all. Considering timing data on the action-level in addition to action sequences, however, has vast potential to support a more fine-grained assessment of examinees' behavior. We provide an approach that jointly considers action sequences and action-level times for identifying common response processes. In doing so, we integrate tools from clickstream analyses and graph-modeled data clustering with psychometrics. In our approach, we (a) provide similarity measures that are based on both actions and the associated action-level timing data and (b) subsequently employ cluster edge deletion for identifying homogeneous, interpretable, well-separated groups of action patterns, each describing a common response process. Guidelines on how to apply the approach are provided. The approach and its utility are illustrated on a complex problem-solving item from PIAAC 2012.

In medical science and practice, dietary behavior is mostly assessed by item-extensive questionnaires (e.g., food-frequency-questionnaires) or by questionnaires focusing on psychological aspects of dietary behavior neglecting dietary quality or quantity. In consequence, these questionnaires do not capture the full bandwidth of dietary behavior or are less effective in the assessment of dietary behavior because of the large item pools. Therefore, the aim of this validation study was to translate the existing General Dietary Behavior Inventory (GDBI), which was constructed as a behavior-related, as well as effective, instrument, and verifying its construct and criterion validity. This inventory is based on the general nutrition recommendations of the World Health Organization (WHO). Our English-speaking convenience sample consisted of 263 participants. The study results confirmed convergent, as well as criterion validity of the English version of the GDBI (GDBI-E). Discriminant validity of the GDBI-E could mainly be verified. Different dietary behavior clusters were identified in a cluster analysis. The found clusters represented a rather healthy and a rather unhealthy dietary behavior in the sample according to the recommendations of the WHO. The results underpinned the validity of the GDBI-E. The GDBI-E is applicable in research and clinical practice to assess dietary behavior in the English-speaking population. OA Förderung 2022

This paper explores the trends in participation in a variety of cultural activities in Slovakia. Although the research on cultural participation has thrived and there have been plenty of empirical studies published about the consumption of cultural products in different countries, there is still a lack of comprehensive research from former post-communist countries. The paper aims to test the link between cultural participation and social and economic characteristics, which has previously been done in other international studies. The cluster analysis suggests four types of Slovak cultural participants&mdash, the heritage visitors, the omnivores, the popular genre visitors, and the inactive. The outcomes point to a strong relationship between cultural inactivity and heritage visits to socioeconomic status, while there is a weak relationship between highbrow cultural participation and the upper-class. The results challenge cultural policies to be adjusted for different groups of cultural participants with the objective of increasing the overall consumption of cultural goods in Slovakia.

Attention towards topics such as environmental pollution, climate change, or biodiversity has strongly increased in the last years. The struggles to balance market powers and ecological sustainability somehow evoke memories of the early days of European welfare states, when social protection emerged as a means to prevent industrial capitalism from disruptive social tensions due to excessive social inequalities. In fact, social and environmental crises are inseparably intertwined, as ecological destruction is likely to be followed by social deprivation, and a lack of social security can be a crucial barrier for ecologically sustainable action. Our paper seeks to provide a step towards such an integrated perspective by studying problem pressure and public interventions in the area of green welfare, that is, in social and environmental protection. By using available data from Eurostat and Environmental Performance Index (EPI) databases, we contrast environmental and social performances to detect links between the social and the ecological dimension in these areas and unearth different configurations of green welfare among European countries. Our findings suggest that there are different “worlds of eco-welfare states” which only partially overlap with the more conventional “world of welfare states” but show how the Nordic countries are in the relatively-better performing cluster.

Die vorliegende Arbeit wurde im Rahmen einer Industriepromotion bei der Daimler AG in Sindelfingen erstellt. Sie umfasst die Entwicklung und Beschreibung eines statistischen Methodenbaukastens um Last- und Verschleißanalysen prozessual durchführen zu können. Dieser Methodenbaukasten wird an Daten im Kontext alternativer Antriebssysteme beispielhaft erprobt. Der methodische Fokus liegt auf der Quantifizierung und Sensitivität der Güte bzw. Unsicherheit auf den einzelnen Analysestufen. Die erste Analysestufe beinhaltet die Identifizierung verschiedener Gruppen in Belastungsdaten, umgesetzt durch Clusterverfahren. Auf der zweiten Analysestufe sollen über verschiedene lineare und nichtlineare Verfahren Vorhersagen für das Verschleißverhalten der identifizierten Gruppen getroffen werden. Auf beiden Stufen soll sowohl die Güte des Verfahrens als auch dessen Sensitivität quantifiziert werden. Im Rahmen der Arbeit werden alle notwendigen statistischen Methoden definiert, die entsprechenden Gütekriterien werden eingeführt. Der Methodenbaukasten beinhaltet einen iterativen Prozess, in dem in jeder Iteration sowohl das Clustering als auch die Prognose durchgeführt wird. So kann zum einen in jedem Schritt die Güte des jeweiligen Verfahrens und zum anderen die Sensitivität der Güte bzw. Unsicherheit der Verfahren/Modelle über mehrere Iterationen quantifiziert und bewertet werden. Der entwickelte, iterative Prozess, integriert in den Algorithmus des Evidence Accumulation Clusterings, bietet dem Anwender entscheidende methodische Vorteile. Zum einen kann in jedem Schritt die Güte und dessen Sensitivität des jeweiligen Verfahrens bewertet werden, zum anderen wird über die gleichzeitige Durchführung aller Verfahren in jeder Iteration beides über die Analysestufen hin weg quantifiziert. Im Anwendungsbeispiel werden Potentiale aufgezeigt, die Güte der Modelle zu steigern sowie die Sensitivität zu verringern, indem sowohl die Variablenselektion für die Lastanalyse als auch die Modellauswahl für die Verschleißprognose prozessual durchgeführt wird. Der entwickelte Prozess bietet die Möglichkeit, die Qualität und Stabilität der durchgeführten Analyse bereits zu frühen Zeitpunkten (geringe Datenbasis) zu bewerten und ggf. Handlungsmaßnahmen abzuleiten.

Diversification is one of the main pillars of investment strategies. The prominent 1/N portfolio, which puts equal weight on each asset is, apart from its simplicity, a method which is hard to outperform in realistic settings, as many studies have shown. However, depending on the number of considered assets, this method can lead to very large portfolios. On the other hand, optimization methods like the mean-variance portfolio suffer from estimation errors, which often destroy the theoretical benefits. We investigate the performance of the equal weight portfolio when using fewer assets. For this we explore different naive portfolios, from selecting the best Sharpe ratio assets to exploiting knowledge about correlation structures using clustering methods. The clustering techniques separate the possible assets into non-overlapping clusters and the assets within a cluster are ordered by their Sharpe ratio. Then the best asset of each portfolio is chosen to be a member of the new portfolio with equal weights, the cluster portfolio. We show that this portfolio inherits the advantages of the 1/N portfolio and can even outperform it empirically. For this we use real data and several simulation models. We prove these findings from a statistical point of view using the framework by DeMiguel, Garlappi and Uppal (2009). Moreover, we show the superiority regarding the Sharpe ratio in a setting, where in each cluster the assets are comonotonic. In addition, we recommend the consideration of a diversification-risk ratio to evaluate the performance of different portfolios. Der fundamentalste Grundgedanke aller Anlagestrategien ist die Diversifikation. Das 1/N-Portfolio, welches in jede Anlage gleich viel investiert, folgt diesem Grundgedanken am meisten. Jedoch führt es dadurch zu einer großen Anzahl von Einzelinvestitionen. Andererseits gibt es Ansätze aufgrund der Parameter der einzelnen Anlagen eine optimale Strategie zu finden. Diese Strategien, zu denen auch die Mean-Variance Optimierung gehört, leiden dabei unter dem Einfluss von Schätzfehlern, die allzu oft die theoretischen Vorteile in der Praxis auslöschen. Wir untersuchen daher die Auswirkungen auf naive Portfolios, wenn man sie auf eine reduzierte Anzahl von Anlagen anwendet. Dazu betrachten wir verschiedene Auswahlkriterien, von der Wahl der besten Anlagen bezüglich des Sharpe-Ratios bis zur Ausnutzung von Korrelationsstrukturen im Markt durch Clusterbildung. Das Clustern teilt die vorhandenen Anlagen in separate Gruppen mit hoher interner Korrelation auf. Dann werden die Anlagen innerhalb der Cluster nach ihrem Sharpe-Ratio sortiert. Das Cluster-Portfolio besteht dann aus den besten Repräsentanten der einzelnen Cluster. Wir können empirisch zeigen, dass das Cluster-Portfolio nicht nur die guten Eigenschaften des naiven Portfolios übernimmt, sondern es auch oft übertrifft. Dies belegen wir mit Studien zu simulierten und realen Daten. Zudem können wir einige Effekte theoretisch erklären. Der Arbeit von DeMiguel, Garlappi und Uppal (2009) folgend tun wir das aus einem statistischen Blickwinkel. Im Spezialfall mit komonotonen Gruppen können wir den besseren Sharpe-Ratio nachweisen. Zusätzlich schlagen wir die Betrachtung eines Diversifikations-Risiko Quotienten vor, der der besseren Beurteilung verschiedener Portfolios dienen soll.

Vaccine hesitancy is not a singular view but encompasses a set of positions located between complete acceptance of vaccination and complete rejection of vaccination. In this paper, I argue that vaccine-hesitant attitudes emerge at the intersection of individual and structural processes, and thus can be better conceptualized as “extended attitudes”. Drawing on the theoretical understanding of risk and science scepticism in post-modern societies, I consider hesitant attitudes towards vaccination as addressing risks that are induced in our everyday lives by science developments. I conducted K-Means Cluster Analysis on Eurobarometer data from 2019 regarding Europeans’ attitudes towards vaccination. Four clusters of vaccine-hesitant attitudes were identified. “Price hesitation” and “Effort hesitation” result from restricted access to vaccination because of structural constraints, such as low economic capital and health care system’ deficits. “Unexercised pro-vaccination” is an attitude manifested by people who grant authority to science to manage health-related risks, even though they did not vaccinate in the last five years. “Consistent anti-vaccination” pertains to highly reflexive individuals who dismiss experts’ authority because of scientifically derived risks. My analysis enhances the theoretical understanding and the empirical assessment of vaccine-hesitant attitudes in the European Union and can inform public health policies in this area.

Die Clusteranalyse ist ein mächtiges Werkzeug um Gruppen von Datenpunkten mit ähnlichen Eigenschaften zu identifizieren. Verschiedene Algorithmen finden diese Gruppen indem sie das k-means Optimierungsproblem lösen. Da die Lösung dieses Problems NP-schwer ist, wurden Heuristiken entwickelt um die Optimierung zu beschleunigen. Eine dieser Heuristiken wurde von Lloyd vorgestellt und hat sich zum Standard k-means Algorithmus entwickelt. Über den Zeitraum von mehreren Dekaden wurden in Forschungsarbeiten zahlreiche Lloyd Varianten vorgestellt mit dem Ziel den originalen Algorithmus zu beschleunigen. Fast alle dieser Arbeiten hatten gemeinsam, dass eine Beschleunigung auf nahezu allen getesteten Datensätzen erreicht werden sollte. Es wird ein Rahmenwerk präsentiert um k-means Varianten zu erzeugen welche lediglich auf Datensätzen mit einer spezifischen Datenrepräsentation effektiv sind. Mit detailliertem Wissen über die Struktur eines Datensatzes ist es somit möglich die schnellste Variante passend zu dem Datensatz auszuwählen. Um die Varianten zu erzeugen werden Lower Bound Feature Maps vorgestellt welche verschiedene Datenrepräsentationen ausnutzen können um Distanzen zu approximieren. Die Approximationen benötigen zur Berechnung nur einen Bruchteil der Zeit einer vollständigen Distanzberechnung and eignen sich daher zur Beschleunigung des k-means Algorithmus. Anhand einer sehr diversen Auswahl von Datensätzen unterschiedlicher Datentypen (Video, Text, ausführbare Dateien) und Spärlichkeit wird gezeigt, dass Varianten mit aktivierter Feature Map Optimierung signifikant schneller sind als nicht optimierte Varianten. Weitergehende Untersuchungen haben gezeigt, dass die Lower-Bounds effektiver werden je höher k gewählt wird. Dies führte zu der Entwicklung von hoch optimieren sehr schnellen Varianten, welche für große k einen Speicherverbrauch ähnlich zu Lloyd’s Algorithmus aufweisen., Cluster analysis is a powerful tool to identify groups of data points with similar properties. Various algorithms find these groups by optimizing the well-known k-means objective. As its optimization was proven to be NP-hard, heuristics were developed to speed up the optimization. One of these heuristics was developed by Lloyd and is used today within the algorithm known as standard k-means. Over decades researchers proposed variants of Lloyd’s algorithm with the goal to be faster than the original. Achieving a maximum speedup on all tested data sets was one of the key properties that the majority of these researchers were focused on. We present a framework for creating k-means variants where every variant is only effective on a data set with a specific representation. With detailed knowledge about a data set one is able to choose the fastest k-means variant for the clustering task at hand which is often faster than traditional algorithms. To create the variants, lower bound feature maps are introduced that exploit different representations and are used to approximate distances. The approximations are tight if a feature map fits to a specific data representation and loose otherwise. Tight approximations, which in comparison to the actual distance can be calculated in a fraction of the time, can be used to speed up the k-means algorithm. On a variety of data sets with different types (video, text, executable files) and sparsities it is demonstrated that feature map optimized variants can achieve considerable speedups over their non feature map optimized counterparts. Moreover, it is shown that these bounds get more effective when k is increased. This leads to variants that have a memory consumption close to Lloyd‘s algorithm when clustering data sets with large k.

Even though the turn to practice is widely accepted in the field of urban planning, the practices of planners are empirically largely unexplored. Looking at the daily routines and practices of urban planners thus allows a deeper insight into what planning is, and how planning practices are related to innovation and transformation. To do so, we start from the assumption that behaviour is a constellation of practices, including certain activities, a set of choices and actions, patterns of behaviour or forms of interaction that is organised in a certain space or context by common understandings and rules. By conducting an online survey among planners in medium-sized German cities, we first identified a wide range of planning practices and activities in general. In a second step, we conducted a statistical cluster analysis resulting in six types of planners: (1) the ‘local-specific analysts,’ (2) the ‘experienced generalists,’ (3) the ‘reactive pragmatists,’ (4) the ‘project-oriented planners,’ (5) the ‘compensatory moderators,’ and (6) the ‘innovative designers.’ Each cluster has specific practices and activities, linked to characteristic value-sets, role interpretations and self-perceptions that might help explain the differences with regard to innovation and transformation. From the identified six groups or clusters of planners, only two clusters more or less consequently aim at innovation, experimentation and new approaches. One cluster is dedicated to collaborative practices whereas traditional practices predominate in three clusters at least, mainly because of legal requirements. This is the result of an increasing ‘formalisation’ of land-use planning, making planners focus on technical and formal practices, and, at the same time, lead to the reduced ‘attention’ to and implementation of conceptual approaches or ‘necessary’ transformative practices, including proactive approaches and strategic coordination with regard to sustainable urban development, but also comprising experiments, real labs or social innovations., Urban planning;Vol 4, No 4 (2019)