Search

In this study, we investigate the impact of data sampling on the stability of uplift models. The aim of uplift modelling is to predict people’s reaction to a campaign offering incentives towards taking a particular action, such as buying a product or not churning from a service. In the literature, it has been reported that uplift models can be unstable performance-wise (in terms of the Qini measure for example) and may consequently be deemed unreliable. We investigate whether model stability can be improved by modifying the class ratio using data sampling techniques, without worsening the performance in the process. Specifically, we use both under- and oversampling, as well as the methods smote and rose. Whereas the first two remove or duplicate records, the latter two artificially generate new records taking account of their neighbourhood while potentially also removing records. The uplift models are built considering uplift random forests, and the methods of Lo and Generalized Lai. In our experimental design, we take account of different class ratios and we use a number of direct marketing datasets. Furthermore, we check whether these new models show an increased performance, studying whether there exists a tradeoff between stability and performance. In some cases, we observe that sampling methods have a positive impact on both the stability and performance of uplift models, but this depends on the method and dataset used.

This study uses social network analysis to investigate the evolution of international trade and the globalization of the economy over the period 2000-2013. We employ a directed and weighted approach to build and analyze the world trade web using the UN Comtrade’s exportation data between all countries. Since visualization of such large networks is not very representative nor useful, our framework looks in detail at centrality measures on a global as well as a local level. Globally, we look mainly at the degree (also its subparts in- and outdegree) and its weighted version, the (in/out) strength. The betweenness and closeness are also analyzed, but mainly for comparison with the local levels. For these local levels, we chose to analyze the centrality measures on the second order neighborhood. This choice is based on the network variables calculated on the global level. This will allow us to find out which countries are used most in trade routes. These are not the same countries as the ones on a global level. On a final note, we analyze the network and behavior dynamics in order to calculate how the network evolves in general and how it evolves based on different criteria (same language, continent etc.).

In this paper, we investigate the global R&D network that emerged from Green Car Technology research by firms and universities. We scrutinize the development of the relevant patent network over time and investigate whether engineered (brokered) collaborations in R&D networks are more stable and/or intensive than collaborations that emerge otherwise. In our analysis, we distinguish between industry-university, inter-industry and inter-university R&D collaborations to uncover whether collaboration with or between universities leads to more stable or unstable collaborations. Insights from our research will allow us to uncover the reasons for collaborative R&D with universities and highlight how Green Car firms access relevant third party knowledge. Furthermore, we will be able to assess the importance of engineered relations and universities in the emergence and development of R&D networks. We control for regional differences in R&D networks and possible learning effects that might occur in the network. Further research will allow for strategic R&D tool development and for scrutinizing the role of individual researchers in the development and evolution of collaborative R&D networks. In literature, similar networks are often scrutinized, yet very rarely these networks are studied as a whole and over their lifecycle.