Scholarlink

<p>Accurate forecasting of agricultural product prices is imperative for ensuring market stability and informing effective policy-making. This paper introduces a novel forecasting framework, termed time series segmentation-based gradient boosting regression trees (TSS-GBRT), designed to leverage the temporal dynamics inherent in agricultural product price data. The model incorporates two unique hyperparameters, periodic unit sampling and phase direction sampling, to enhance its ability to capture seasonal patterns while maintaining learning efficiency. Empirical validation using Japanese agricultural product price data demonstrates a significant enhancement in forecasting performance, achieving a 5.84% reduction in error compared to baseline methods, along with a 24-min reduction in learning time. The findings underscore the potential of TSS-GBRT as a lightweight, interpretable, and highly effective tool for addressing the complexities of agricultural product price forecasting.</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

Development of an agricultural product price forecasting model via gradient boosting regression trees with a time series sampling approach

  • 期刊论文

  • 发表日期：

    2025年7月31日

查看全文
  • Kazuki Fujiwara

摘要

Accurate forecasting of agricultural product prices is imperative for ensuring market stability and informing effective policy-making. This paper introduces a novel forecasting framework, termed time series segmentation-based gradient boosting regression trees (TSS-GBRT), designed to leverage the temporal dynamics inherent in agricultural product price data. The model incorporates two unique hyperparameters, periodic unit sampling and phase direction sampling, to enhance its ability to capture seasonal patterns while maintaining learning efficiency. Empirical validation using Japanese agricultural product price data demonstrates a significant enhancement in forecasting performance, achieving a 5.84% reduction in error compared to baseline methods, along with a 24-min reduction in learning time. The findings underscore the potential of TSS-GBRT as a lightweight, interpretable, and highly effective tool for addressing the complexities of agricultural product price forecasting.