By Guanzhou Wei, Xiao Liu, and Russell R. Barton📧
In Environmetrics, 2023, e2831, early view online version, published online: October 26. https://doi.org/10.1002/env.2831
Partial differential equation (PDE)-based spatio-temporal models are available in the literature for modeling spatio-temporal processes governed by advection-diffusion equations. The main idea is to approximate the process by a truncated Fourier series and model the temporal evolution of the spectral coefficients by a stochastic process whose parametric structure is determined by the governing PDE. However, because many spatio-temporal processes are nonperiodic with boundary discontinuities, the truncation of Fourier series leads to the well-known Gibbs phenomenon (GP) in the output generated by the existing PDE-based approaches. This article shows that the existing PDE-based approach can be extended to suppress GP. The proposed approach starts with a data flipping procedure for the process respectively along the horizontal and vertical directions, as if we were unfolding a piece of paper folded twice along the two directions. For the flipped process, this article extends the existing PDE-based spatio-temporal model by obtaining the new temporal dynamics of the spectral coefficients. Because the flipped process is spatially periodic and has a complete waveform without boundary discontinuities, GP is removed even if the Fourier series is truncated. Numerical investigations show that the extended approach improves the modeling and prediction accuracy. Computer code is made available on GitHub.
Keywords: Advection-diffusion processes; Gibbs phenomenon suppression; PDE-based statistical learning; Spatial-temporal models
Graph anomaly detection is attracting remarkable multidisciplinary research interests ranging from finance, healthcare, and social network analysis. Recent advances on graph neural networks have substantially improved the detection performance via semi-supervised representation learning. However, prior work suggests that deep graph-based methods tend to learn spurious correlations. As a result, they fail to generalize beyond training data distribution. In this article, we aim to identify structural and contextual anomaly nodes in an attributed graph. Based on our preliminary data analyses, spurious correlations can be eliminated with causal subgraph interventions. Therefore, we propose a new graph-based anomaly detection model that can learn causal relations for anomaly detection while generalizing to new environments. To handle situations with varying environments, we steer the generative model to manufacture synthetic environment features, which are exerted on realistic subgraphs to generate counterfactual subgraphs. Further, these counterfactual subgraphs help a few-shot anomaly detection model learn transferable and causal relations across different environments. The experiments on three real-world attributed graphs show that the proposed approach achieves the best performance compared to the state-of-the-art baselines and learns robust causal representations resistant to noises and spurious correlations.
Industry 4.0 integrates digital and physical technologies to transform work management, where two core enablers are the Internet of Things (IoT) and 3D printing (3DP). IoT monitors complex systems in real-time, while 3DP enables agile manufacturing that can respond to real-time information. However, the details of how these two can be integrated are not yet clear. To gain insights, we consider a scenario where a 3D printer supplies a critical part to multiple machines that are embedded with sensors and connected through IoT. While the public perception indicates that this integration would enable on-demand printing, our research suggests this is not necessarily the case. Instead, the true benefit is the ability to print predictively. In particular, it is typically more effective for the 3D printer to predictively print-to-stock, based on a threshold that depends on the system’s status. We also identify a printing mode called predictive print-on-demand that allows for minimal inventory, and find the speed of 3DP to be the primary factor that influences its optimality. Furthermore, we assess the value of IoT in cost reductions by separately analyzing the impact of advance information from embedded sensors and the real-time information fusion through IoT. We find that IoT provides significant value in general. However, the conventional wisdom that IoT’s value scales up for larger systems is suitable only when the expansion is paired with appropriate 3DP capacity. Our framework can help inform investment decisions regarding IoT/embedded sensors and support the development of scheduling tools for predictive 3D printing.
Mutual funds are becoming increasingly popular with the emergence of Internet finance. Clear profiling of a fund’s investment style is crucial for fund managers to evaluate their investment strategies, and for investors to understand their investment. However, it is challenging to profile a fund’s investment style as it requires a comprehensive analysis of complex multi-dimensional temporal data. In addition, different fund managers and investors have different focuses when analysing a fund’s investment style. To address the issue, we propose iFUNDit, an interactive visual analytic system for fund investment style analysis. The system decomposes a fund’s critical features into performance attributes and investment style factors, and visualizes them in a set of coupled views: a fund and manager view, to delineate the distribution of funds’ and managers’ critical attributes on the market; a cluster view, to show the similarity of investment styles between different funds; and a detail view, to analyse the evolution of fund investment style. The system provides a holistic overview of fund data and facilitates a streamlined analysis of investment style at both the fund and the manager level. The effectiveness and usability of the system are demonstrated through domain expert interviews and case studies by using a real mutual fund dataset.
Content providers in online social media platforms, particularly livestreaming, often switch content categories. Despite its uniqueness and importance, there is a dearth of academic research examining the unintended effects of providers’ content switching. We study the direct and indirect spillover effects of content switching for livestreamers—individuals who broadcast content through livestreaming platforms. We propose a framework based on theories related to viewer flow and network effects to conceptualize the direct and indirect spillover effects of entrant streamers’ content switching on the incumbent streamers. Contrary to conventional wisdom, which concerns the negative effects on the incumbent’s viewership, we propose two positive spillover effects that are unique to the social media platform setting: (a) the entrant streamers do not just increase competition among streamers, but they also bring their own viewers to the new category, which benefits the incumbent streamers because of a streaming flow effect (direct spillover), and (b) the entrant streamers influence incumbent streamers’ viewer size by boosting category visibility through indirect network effects (indirect spillover). We also propose that the two spillover effects are contingent on the size of the entrant streamers’ follower base. Based on a unique observational data set from the leading livestreaming platform (Twitch.tv), particularly with viewer flow data at the streamer–session level, we first estimate that average content switching is associated with a 1.3% net increase in direct net viewer flow from the entrant to an incumbent. And this direct spillover effect is attenuated by the size of the entrant streamers’ follower base. We also estimate that average content switching is associated with a 2.6% net increase in (indirect) net viewer flow from outside categories to an incumbent streamer. And this indirect spillover effect is reinforced by the entrant streamers’ follower base size. This study contributes to the emerging literature on the dynamics of content creation on social media platforms in the emerging context of livestreaming. We discuss the managerial implications of this study for streaming strategies and platform management.
Businesses have a range of choices to make among different manufacturing modes (methods/processes), production environment, and planning strategies, depending on the nature of the product and the target market. 
In our globalized economy, multinational corporations (MNCs) often work with multiple tiers of suppliers, many of them located in developing nations with poor infrastructure and fewer government regulations. The multi-tiered nature of modern supply networks creates a complex network of producers responsible for the goods of a single firm which makes tracking production and managing disruption exceedingly difficult. Adding to that challenge is the increasing demand by the MNCs’ customers for ethically sourced products and a desire to understand where and how an item is made. With each passing year and especially in the face of COVID-19 pandemic and its multitude of challenges, MNCs have made efforts to improve practices and meet ever changing environmental, social, and governance (ESG) standards and expectations. Common practices for ESG standards implementation, where procurement takes a central role in communicating and working with the firms’ suppliers to achieve these standards, have been well documented in the literature. These common practices that have been shown to be effective in affecting change in supplier behavior, however, are not often considered when an ESG rating organization confers its grades. By outlining the current practices and detailing several common ESG rating indices, this paper highlights the misalignment between ESG procurement practices and the rating indices that reward firms for their ESG efforts.
Contemporary supply chains have become more comprehensive as a result of economic globalization, and increased product and service complexity driven by ever-increasing consumer demand and expectations. Rising against this backdrop is the importance of supply chain compression. While compression-based exploits can be pursued across supply chain processes, doing so in the new product development (NPD) stage can provide a great opportunity to realize the value of compression strategies. An emerging trend in the strive for rapid innovation is the transition from traditional innovation models that were local, physical, sequential, and product-centric to one that is more global, more virtual, more concurrent and iterate, and more customer- and experience-centric. Under the new paradigm, 3D virtual prototyping (VP) technologies—enhanced by the advancement of extended reality (XR) techniques that includes virtual reality (VR), augmented reality (AR), and mixed reality (MR)—have garnered attention in various industries for their potentials to enable NPD compressions. Despite its promises, VP technology applications under the new NPD model are still nascent and the technology itself continues to evolve. To offer a better understanding in this area, this white paper introduces the concept of supply chain compression (SCC), provides an overview of the fundamental differences between conventional and new concurrent NPD models, and discusses potential applications of XR-enhanced VP techniques in NPD activities and resulting compression-induced business values.