Scholarlink

<p>The precise prediction of investment decisions by returning Huizhou merchants in Anhui Province and the optimization of related policies are critical issues in current regional governance, given their role as the core driving force behind high-quality regional economic development. The present study proposes a novel model based on decision transformers and multimodal data fusion to predict and optimise the investment intentions of new Huizhou merchants returning to Anhui Province. Conventional models, such as LSTM, encounter challenges in accounting for nonlinear correlations and dynamic policy responses, resulting in substantial prediction errors and suboptimal resource allocation. The model under discussion successfully circumvents the limitations of traditional methods by means of the integration of spatiotemporal attention mechanisms. These mechanisms are employed for the purpose of analysing technical indicators, policy texts and social capital. The enhanced NSGA-II algorithm, when employed in conjunction with the q-power gradient update strategy, facilitates the optimisation of dynamic policies, thereby ensuring a balanced consideration of fiscal constraints, technology spillover effects, and investment risks. Empirical evidence has demonstrated that prediction errors in high-tech industries are reduced by 46%, policy response delays are decreased by 62%, and technology conversion rates reach 74.6%. This study validates the effectiveness of the "technology-policy-capital" collaborative governance framework, providing a robust solution for regional economic development and policy formulation under the Yangtze River Delta integration strategy. Future research will focus on model lightweighting, cross-regional policy simulation, and the optimization of real-time risk response mechanisms.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on the prediction and policy optimization model of Anhui merchants' return investment based on decision transformer

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月8日

查看全文
  • Yongqing Zhou

摘要

The precise prediction of investment decisions by returning Huizhou merchants in Anhui Province and the optimization of related policies are critical issues in current regional governance, given their role as the core driving force behind high-quality regional economic development. The present study proposes a novel model based on decision transformers and multimodal data fusion to predict and optimise the investment intentions of new Huizhou merchants returning to Anhui Province. Conventional models, such as LSTM, encounter challenges in accounting for nonlinear correlations and dynamic policy responses, resulting in substantial prediction errors and suboptimal resource allocation. The model under discussion successfully circumvents the limitations of traditional methods by means of the integration of spatiotemporal attention mechanisms. These mechanisms are employed for the purpose of analysing technical indicators, policy texts and social capital. The enhanced NSGA-II algorithm, when employed in conjunction with the q-power gradient update strategy, facilitates the optimisation of dynamic policies, thereby ensuring a balanced consideration of fiscal constraints, technology spillover effects, and investment risks. Empirical evidence has demonstrated that prediction errors in high-tech industries are reduced by 46%, policy response delays are decreased by 62%, and technology conversion rates reach 74.6%. This study validates the effectiveness of the "technology-policy-capital" collaborative governance framework, providing a robust solution for regional economic development and policy formulation under the Yangtze River Delta integration strategy. Future research will focus on model lightweighting, cross-regional policy simulation, and the optimization of real-time risk response mechanisms.