Scholarlink

<p>Traditional Earned Value Management (EVM) is one of the most widely adopted project-control methodologies; however, its application is fundamentally constrained during the conceptual stage because reliable cost, schedule, variance, and risk assessments cannot be generated before actual project progress data becomes available. This paper introduces Automated Conceptual Earned Value Management (ACEVM), a novel project-control concept, methodology, and decision-support framework that extends EVM from a retrospective performance-monitoring tool into a predictive, proactive and intelligent project-control system. ACEVM enables the automated estimation of Budget at Completion, Duration at Completion, Cost Variance, Schedule Variance, and Conceptual Project Risk using limited early-stage project information before project execution begins. The framework was developed and validated using 1,277 completed Field Canal Improvement Projects (FCIPs). Twenty-five predictors were screened through factor analysis and stepwise regression and used to train 60 optimized machine-learning model instances, defined as 12 candidate algorithms evaluated across five ACEVM prediction outputs. Hyperparameters were optimized using the Artificial Bee Colony algorithm, while model evaluation considered predictive accuracy, robustness, explainability, and computational CO<sub>2</sub> emissions. The best-performing models were LightGBM for budget prediction, Extremely Randomized Trees for duration and cost variance prediction, AdaBoost for duration variance prediction, and Extremely Randomized Trees for risk classification.</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

Automated conceptual earned value management

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月25日

查看全文
  • Haytham Elmousalami,
  • Abdulaziz Alotaibi

摘要

Traditional Earned Value Management (EVM) is one of the most widely adopted project-control methodologies; however, its application is fundamentally constrained during the conceptual stage because reliable cost, schedule, variance, and risk assessments cannot be generated before actual project progress data becomes available. This paper introduces Automated Conceptual Earned Value Management (ACEVM), a novel project-control concept, methodology, and decision-support framework that extends EVM from a retrospective performance-monitoring tool into a predictive, proactive and intelligent project-control system. ACEVM enables the automated estimation of Budget at Completion, Duration at Completion, Cost Variance, Schedule Variance, and Conceptual Project Risk using limited early-stage project information before project execution begins. The framework was developed and validated using 1,277 completed Field Canal Improvement Projects (FCIPs). Twenty-five predictors were screened through factor analysis and stepwise regression and used to train 60 optimized machine-learning model instances, defined as 12 candidate algorithms evaluated across five ACEVM prediction outputs. Hyperparameters were optimized using the Artificial Bee Colony algorithm, while model evaluation considered predictive accuracy, robustness, explainability, and computational CO2 emissions. The best-performing models were LightGBM for budget prediction, Extremely Randomized Trees for duration and cost variance prediction, AdaBoost for duration variance prediction, and Extremely Randomized Trees for risk classification.