Stacking多模型融合优化高校图书采购预测的研究
Research on Stacking multiple models for optimizing university library book procurement forecasting
罗可 1阳志花 1陈玫瑰1
作者信息
摘要
提出了一种基于Stacking多模型融合的图书采购预测模型,旨在提升高校图书采购预测的准确性和可靠性.传统的单一预测模型难以较好地应对高校图书采购中的诸多复杂因素.采用Stacking方法,构建了一个次级模型,能够有效整合不同基础模型的预测结果,并通过交叉验证来选择最佳的Stacking模型,以确保模型的稳定性和泛化能力.实验结果表明,Stacking多模型融合方法显著提升了高校图书采购预测的准确性和鲁棒性.这为高校图书采购管理提供了一种有效的决策工具,有望改善资源分配,降低不必要的成本,并提高管理决策的科学性.
Abstract
This paper presents a book procurement forecasting model based on Stacking ensemble of multiple models,aiming to enhance the accuracy and reliability of book procurement forecasts in universities.Traditional single prediction models struggle to effectively account for the various complex factors in university book procurement.By employing the Stacking method,a second-ary model is constructed to efficiently integrate predictions from different base models,and the best Stacking model is chosen through cross-validation to ensure model stability and generalization.Experimental results demonstrate that the Stacking ensemble of multiple models significantly improves the accuracy and robustness of book procurement forecasts in universities.This offers an effective decision-making tool for university book procurement management,with the potential to enhance resource allocation,re-duce unnecessary costs,and elevate the scientific rigor of management decisions.
关键词
Stacking集成算法/LightGBM/图书采购预测/资源分配Key words
stacking ensemble algorithm/LightGBM/book procurement forecasting/resource allocation引用本文复制引用
基金项目
湖南省哲学社会科学基金项目(21YBA179)
出版年
2024
NETL
NSTL
万方数据