Scholarlink

<p>As human capital forms the backbone of any organization, managing and minimizing employee attrition is of paramount importance. Attrition prediction is essential because attrition disrupts projects advancement, increases rehiring and training costs, and risks losing core knowledge and technologies. Thus, minimizing attrition is essential for organizational stability and competitiveness. In this research, we design a Machine Learning (ML)-based predictive system using six ML models namely Logistic Regression, Random Forest, Gradient Boosting (GB), Decision Tree, Support Vector Machine, and K-Nearest Neighbors and evaluate their performances using accuracy, precision, recall, and f1 score. This system predicts employee attrition in advance, allowing organizations to implement proactive talent management strategies instead of relying on traditional reactive approaches. We analyse ‘IBM HR Analytics data’, that contains 1,470 observations and 35 features. Among all the models, GB outperform others with 98% accuracy, and precision, 100% recall and 99% f1 score. Furthermore, using Explainable AI (XAI) methods such as ELI5, SHAP, LIME, and SHAPASH, we analyze the most influencing factors related to employee attrition. XAI analysis show that ‘Over Time’ ‘Toal working years’, ‘job level’ are the most significant factors. Local explainability provide the low level explanation to reveal the inner story and deep insight.</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

An Explainable Machine Learning-Based Employee Attrition Predictive System

  • 期刊论文

  • 发表日期：

    2025年6月3日

查看全文
  • Amir Hamja,
  • Mahmudul Hasan,
  • Ziaul Hassan,
  • Md Palash Uddin,
  • Md Jahangir Alam Siddikee

摘要

As human capital forms the backbone of any organization, managing and minimizing employee attrition is of paramount importance. Attrition prediction is essential because attrition disrupts projects advancement, increases rehiring and training costs, and risks losing core knowledge and technologies. Thus, minimizing attrition is essential for organizational stability and competitiveness. In this research, we design a Machine Learning (ML)-based predictive system using six ML models namely Logistic Regression, Random Forest, Gradient Boosting (GB), Decision Tree, Support Vector Machine, and K-Nearest Neighbors and evaluate their performances using accuracy, precision, recall, and f1 score. This system predicts employee attrition in advance, allowing organizations to implement proactive talent management strategies instead of relying on traditional reactive approaches. We analyse ‘IBM HR Analytics data’, that contains 1,470 observations and 35 features. Among all the models, GB outperform others with 98% accuracy, and precision, 100% recall and 99% f1 score. Furthermore, using Explainable AI (XAI) methods such as ELI5, SHAP, LIME, and SHAPASH, we analyze the most influencing factors related to employee attrition. XAI analysis show that ‘Over Time’ ‘Toal working years’, ‘job level’ are the most significant factors. Local explainability provide the low level explanation to reveal the inner story and deep insight.