Search

With the rapid development of the mobile Internet, online social networks are playing an increasingly vital role in the dissemination of information. Accurately predicting the size of information cascades in advance has become a crucial issue, particularly in the realms of viral marketing, risk management and resource allocation. There are numerous studies that have tackled this prediction task, but the outcomes are unsatisfactory. In this paper, we explore the predictability of information cascade size through the lens of percolation theory. Our investigation reveals that the accuracy of cascade size prediction is notably diminished in the proximity of the threshold, evident in both artificial and empirical networks. Moreover, we observe a degradation and an user-level difference in prediction performance as social media platforms undergo evolution. Our findings underscore the necessity for additional factors to enhance prediction accuracy. [ABSTRACT FROM AUTHOR]

Ocean memory is crucial for improving climate models and enhancing the accuracy of climate predictions. Despite its importance, the evolving influence of ocean memory over recent decades on monsoon predictions remains ambiguous. The persistence of sea surface temperature (SST) anomalies, a critical indicator of ocean memory, governs local air-sea coupling processes affecting the Asian-Australian monsoon (A-AM), thereby significantly shaping climate predictions for Asia, Australia, and the broader Indo-Pacific region. Drawing on observational and numerical modeling evidence, the study elucidates that within the context of interdecadal variation in ocean memory, the seasonal persistence of Maritime Continent (MC) SST anomalies was more pronounced during the epoch of robust ocean memory (1982-1999). This persistence sustained the anomalous western North Pacific anticyclone (WNPAC) through an intensified Matsuno-Gill response in the decaying phase of El Niño-Southern Oscillation (ENSO), thereby enhancing its linkage with the A-AM system throughout the monsoon year, contributing to improved prediction skills of the A-AM. In contrast, these air-sea coupling processes have weakened during the weak memory epoch (2000-2017), making it more difficult to capture the characteristics of the A-AM. The decline in MC ocean memory at the onset of the twenty-first century has undermined the prediction skills of the leading modes of the A-AM. Overall, this study underscores the profound influence of ocean memory on monsoon prediction skills while highlighting the significant challenges in addressing A-AM variability and predictability against the backdrop of global warming. [ABSTRACT FROM AUTHOR]

Conducting predictability studies is essential for tracing the source of forecast errors, which not only leads to the improvement of observation and forecasting systems, but also enhances the understanding of weather and climate phenomena. In the past few decades, dynamical numerical models have been the primary tools for predictability studies, achieving significant progress. Nowadays, with the advances in artificial intelligence (AI) techniques and accumulations of vast meteorological data, modeling weather and climate events using modern data-driven approaches is becoming trendy, where FourCastNet, Pangu-Weather, and GraphCast are successful pioneers. In this perspective article, we suggest AI models should not be limited to forecasting but be expanded to predictability studies, leveraging AI's advantages of high efficiency and self-contained optimization modules. To this end, we first remark that AI models should possess high simulation capability with fine spatiotemporal resolution for two kinds of predictability studies. AI models with high simulation capabilities comparable to numerical models can be considered to provide solutions to partial differential equations in a data-driven way. Then, we highlight several specific predictability issues with well-determined nonlinear optimization formulizations, which can be well-studied using AI models, holding significant scientific value. In addition, we advocate for the incorporation of AI models into the synergistic cycle of the cognition–observation–model paradigm. Comprehensive predictability studies have the potential to transform "big data" to "big and better data" and shift the focus from "AI for forecasts" to "AI for science", ultimately advancing the development of the atmospheric and oceanic sciences. [ABSTRACT FROM AUTHOR]

This paper evaluates stock return predictability with inflation and output gap, which typically enter the Federal Reserve Bank's interest rate setting rule. We introduce Taylor rule fundamentals into the Fed model that relates stock returns to earnings and long‐term yields. Using real‐time data from 1970 to 2008, we find evidence that the Fed model with Taylor rule fundamentals performs better in‐sample and out‐of‐sample than the constant return and original Fed models. Economic significance tests indicate that the models with Taylor rule fundamentals consistently produce higher utility gains than the benchmark models. Though the performance of the Taylor rule model weakens when we extend the sample to include the post‐2008 period characterized by prolonged zero lower bound episodes, it still outperforms the benchmark models. [ABSTRACT FROM AUTHOR]