Scholarlink

Employee attrition occurs when workers leave a firm for whatever reason (such as retirement or a new position) and are not promptly replaced. It is not enough for a firm to recruit the best and brightest; it must also be able to hold on to them. Therefore, it is important to examine a dataset containing information about a firm's employee list in an effort to discover patterns that may give helpful information in understanding why people leave the organization. In this study, we train and test five machine learning models using the IBM attrition dataset. These models include the decision tree, AdaBoost, random forest, XGBoost, and gradient boosting. The first stage in this work was to process the data, which included filling in missing values and giving proper names to categorical variables that had been given numerical labels. We then utilize Plotly Lib to visualize the data, which leads us to insights and allows us to see trends in employee turnover. We have encoded variables, rescaled features, and dealt with unbalanced classes in preparation for developing a classification model by separating the data into training and testing sets. Finally, we evaluated the models again in an effort to optimize the hyperparameters and get better results.

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

A Framework for Prediction of Employee Attrition Using Machine Learning Models on IBM HR Dataset

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • G. Prathiba,
  • Nagaratna P. Hegde

摘要

Employee attrition occurs when workers leave a firm for whatever reason (such as retirement or a new position) and are not promptly replaced. It is not enough for a firm to recruit the best and brightest; it must also be able to hold on to them. Therefore, it is important to examine a dataset containing information about a firm's employee list in an effort to discover patterns that may give helpful information in understanding why people leave the organization. In this study, we train and test five machine learning models using the IBM attrition dataset. These models include the decision tree, AdaBoost, random forest, XGBoost, and gradient boosting. The first stage in this work was to process the data, which included filling in missing values and giving proper names to categorical variables that had been given numerical labels. We then utilize Plotly Lib to visualize the data, which leads us to insights and allows us to see trends in employee turnover. We have encoded variables, rescaled features, and dealt with unbalanced classes in preparation for developing a classification model by separating the data into training and testing sets. Finally, we evaluated the models again in an effort to optimize the hyperparameters and get better results.