《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (8): 2541-2547.DOI: 10.11772/j.issn.1001-9081.2025070854

• 先进计算 • 上一篇    下一篇

基于树基模型的双阶段能耗特征选择方法

田博然1, 周建涛1,2,3,4(), 赵大明5   

  1. 1.内蒙古大学 计算机学院,呼和浩特 010021
    2.生态大数据教育部工程研究中心(内蒙古大学),呼和浩特 010021
    3.内蒙古自治区云计算与服务软件工程实验室(内蒙古大学),呼和浩特 010021
    4.大数据分析技术内蒙古自治区工程实验室(内蒙古大学),呼和浩特 010021
    5.清华大学 计算机科学与技术系,北京 100084
  • 收稿日期:2025-07-29 修回日期:2025-09-22 接受日期:2025-09-22 发布日期:2025-11-05 出版日期:2026-08-10
  • 通讯作者: 周建涛
  • 作者简介:田博然(2000—),女,河北廊坊人,硕士研究生,主要研究方向:云计算节能资源调度
    周建涛(1974—),女,河北承德人,教授,博士,CCF会员,主要研究方向:形式化方法、云计算、软件工程
    赵大明(1993—),男,黑龙江哈尔滨人,博士,主要研究方向:云计算节能资源调度、大语言模型服务优化。
  • 基金资助:
    国家自然科学基金资助项目(62162046);国家自然科学基金资助项目(62472236);呼和浩特市重大科技专项(23-战-重-1)

Two-stage energy consumption feature selection method based on tree-based models

Boran TIAN1, Jiantao ZHOU1,2,3,4(), Daming ZHAO5   

  1. 1.College of Computer Science,Inner Mongolia University,Hohhot Inner Mongolia 010021,China
    2.Engineering Research Center of Ecological Big Data (Inner Mongolia University),Ministry of Education,Hohhot Inner Mongolia 010021,China
    3.Inner Mongolia Engineering Laboratory for Cloud Computing and Service Software (Inner Mongolia University),Hohhot Inner Mongolia 010021,China
    4.Inner Mongolia Engineering Laboratory for Big Data Analysis Technology (Inner Mongolia University),Hohhot Inner Mongolia 010021,China
    5.Department of Computer Science and Technology,Tsinghua University,Beijing 100084,China
  • Received:2025-07-29 Revised:2025-09-22 Accepted:2025-09-22 Online:2025-11-05 Published:2026-08-10
  • Contact: Jiantao ZHOU
  • About author:TIAN Boran, born in 2000, M. S. candidate. Her research interests include energy efficiency and resource scheduling in cloud computing, resource scheduling.
    ZHAO Daming, born in 1993, Ph. D. His research interests include energy efficiency and resource scheduling in cloud computing, resource scheduling, large language model service optimization.
  • Supported by:
    National Natural Science Foundation of China(62162046);Major Science and Technology Special Project of Hohhot City(23-Zhan-Zhong-1)

摘要:

云计算平台中应用服务的多维资源特征显著影响数据中心能耗。通过特征选择技术提取关键能耗指标是构建精准预测模型的有效手段,当前研究未能充分融合特征的可解释性与全局最优的搜索能力,导致所选特征存在冗余度高、预测精度差或决策不可解释等问题。为此,提出一种基于树基模型的双阶段能耗特征选择方法:在第一阶段采用基于博弈论的树基模型可解释技术(TreeSHAP)量化特征的边际贡献,实现高透明度的冗余特征剔除;在第二阶段融合蚁群优化(ACO)算法的快速收敛性与引力搜索算法(GSA)的全局搜索能力,在精简特征空间中定位最优组合。通过两阶段的协同合作,TreeSHAP筛选出的低维可解释空间为ACO-GSA奠定优化基础,大幅降低计算复杂度,ACO-GSA则在此基础上通过混合搜索策略,高效辨识出关键特征组合。通过对比不同特征选择方法,并在多种预测模型上验证特征子集的性能,从特征维度、预测精度及泛化性能等方面进行综合评价。实验结果表明,所提方法在墨尔本大学云数据集上获得的特征子集维度相较于最小绝对收缩和选择(Lasso)算法降低了58.3%,特征子集预测精度较ACO提升了9.1%。所提方法在特征精简性、模型预测精度和泛化可靠性方面均优于对比方法,验证了所提方法的有效性。

关键词: 特征选择, 能耗预测, Shapley值, 蚁群优化算法, 引力搜索算法

Abstract:

In cloud computing platforms, multidimensional resource features of application services influence data center energy consumption significantly. Extracting key energy consumption indicators through feature selection techniques serves as an effective means to build accurate prediction models. The existing studies fail to adequately integrate feature interpretability with global optimal search capability, resulting in issues such as high redundancy, low prediction accuracy, or lack of decision interpretability in selected features. To address these problems, a two-stage energy consumption feature selection method based on tree-based models was proposed. In the first stage, SHapley Additive exPlanations for Tree-based models (TreeSHAP) was used to quantify marginal contribution of features and realize highly transparent elimination of redundant features. In the second stage, rapid convergence of Ant Colony Optimization (ACO) algorithm was combined with global search capability of Gravitational Search Algorithm (GSA) to localize the optimal feature combination in the reduced feature space. Through the two stages working collaboratively: the low-dimensional interpretable space filtered by TreeSHAP was used to provide a foundation for ACO-GSA, with computational complexity reduced significantly, while key feature combinations were identified by ACO-GSA efficiently through a hybrid search strategy. By comparing different feature selection methods and validating the performance of the feature subsets on multiple prediction models, a comprehensive evaluation was conducted in terms of feature dimensionality, prediction accuracy, and generalization performance. Experimental results demonstrate that the proposed method reduces the feature subset dimensionality by 58.3% compared to Least Absolute Shrinkage and Selection Operator (Lasso) and improves the prediction accuracy of the feature subset by 9.1% compared to ACO on the University of Melbourne cloud dataset. It can be seen that the proposed method outperforms comparative methods in feature compactness, prediction accuracy, and generalization reliability, verifying its effectiveness.

Key words: feature selection, energy consumption prediction, Shapley value, Ant Colony Optimization (ACO) Algorithm, Gravitational Search Algorithm (GSA)

中图分类号: