Explore the vast range of titles available.

As the financial market becomes increasingly complex, stock prediction and anomaly data detection have emerged as crucial tasks in financial risk management. However, existing methods exhibit significant limitations in handling the intricate relationships between stocks and addressing anomalous data. This paper proposes the STAGE framework, which integrates the Graph Attention Network (GAT), Variational Autoencoder (VAE), and Sparse Spatiotemporal Convolutional Network (STCN), to enhance the accuracy of stock prediction and the robustness of anomaly data detection. Experimental results show that the complete STAGE framework achieved an accuracy of 85% after 20 training epochs, which is 10% to 20% higher than models with key algorithms removed. In the anomaly detection task, the STAGE framework further improved the accuracy to 95%, demonstrating fast convergence and stability. This framework offers an innovative solution for stock prediction, adapting to the complex dynamics of real-world markets.

This study utilizes stochastic boosted regression trees (BRT) to investigate the effects of the COVID-19 Movement Control Order (MCO) on air quality in Ipoh City, Malaysia. The model aims to explore the Strength of Interaction Effects (SIE) and Relative Variable Importance (RVI) of key pollutants and meteorological variables impacting PM2.5 concentrations. Hourly data on gaseous pollutants (SO₂, NO₂, CO, O₃) and meteorological conditions (wind direction, wind speed, relative humidity, and temperature) were obtained from the Department of Environment for the periods of January to June in both 2019 and 2020, resulting in 2,231 data points. The BRT model was constructed using R software, with the optimal number of trees (nt = 4,372) determined through Out-of-Bag (OOB) iterations. Model performance was evaluated using various statistical metrics, including a Factor of Two (FAC2) of 0.91, R² values exceeding 0.56 (R = 0.74), and an Index of Agreement (IOA) of 0.67, indicating the model’s robustness. The analysis revealed significant differences in the RVI during the MCO and non-MCO periods. In non-MCO data, PM2.5 concentrations were primarily influenced by CO (18.9%), SO₂ (14.6%), O₃ (12.9%), and wind direction (10.66%). During the MCO, the most important variables were CO (22.6%), RH (13.4%), SO₂ (14.7%), and O₃ (12.1%). Additionally, the SIE analysis highlighted interactions such as CO-wind direction (0.24), O₃-wind speed (0.19), and NO₂-CO (0.15). These findings demonstrate that the BRT model effectively captures the key factors influencing air pollution and their interactions. The results provide valuable insights for urban planners and local authorities, helping them design strategies to mitigate pollutant levels by addressing the most impactful variables. The model could guide policy decisions and optimize air quality management, particularly during periods of reduced human activity or emergency conditions like the MCO.

This study presents a longitudinal model for investigating cloud ERP adoption. Previous work on technology adoption mostly has investigated adoption by looking at one specific phase of adoption. However, we argue that the cross-sectional research design does not sufficiently represent the complexity and the highly volatile process of adoption as a whole. Based on an archival analysis of eighty-seven firm level technology adoption studies, we identified 24 transition factors (TF) contributing to the adoption of information technology. Building on Rogers’ Diffusion of Innovations theory (1995) this research attempts to explore which transition factors are relevant in the distinct phases of cloud ERP adoption. In our work, we model the transition factors as “triggers”, where the desirable outcomes of the transition factor within a specific phase of adoption, will lead to the next stage adoption. The contribution of the paper is twofold. First, we make a context-specific contribution by extracting the different factors which have been studied in the context of cloud ERP systems and categorizing them according to the Diffusion of Innovations theory. Second, we also make a theoretical contribution by employing a longitudinal research design.