Scholarlink

This study investigates the application of seven supervised machine learning (ML) models for predicting employee salaries based on demographic, educational, and occupational attributes. The dataset comprises both continuous and high-cardinality categorical features, for which suitable encoding was applied to facilitate model training. The evaluated models include Linear and Ridge Regression (LinR, RidgeR), Decision Trees (DT), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Gradient Boosting (GB), and Random Forest (RF). Performance is measured using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared ( \( R^2 \) ) metrics. Results show that ensemble models, particularly RF and GB, achieve the highest accuracy, with RF reaching an \( R^2 \) of 0.94. In contrast, kernel and instance-based methods underperform due to limitations in handling categorical data. The findings support the integration of ensemble models into human resource (HR) analytics systems and highlight the importance of model selection based on data structure.

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

Comparative Analysis of Machine Learning Models for Employee Salary Prediction

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Elias Dritsas,
  • Maria Trigka,
  • Phivos Mylonas

摘要

This study investigates the application of seven supervised machine learning (ML) models for predicting employee salaries based on demographic, educational, and occupational attributes. The dataset comprises both continuous and high-cardinality categorical features, for which suitable encoding was applied to facilitate model training. The evaluated models include Linear and Ridge Regression (LinR, RidgeR), Decision Trees (DT), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), Gradient Boosting (GB), and Random Forest (RF). Performance is measured using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R-squared ( \( R^2 \) ) metrics. Results show that ensemble models, particularly RF and GB, achieve the highest accuracy, with RF reaching an \( R^2 \) of 0.94. In contrast, kernel and instance-based methods underperform due to limitations in handling categorical data. The findings support the integration of ensemble models into human resource (HR) analytics systems and highlight the importance of model selection based on data structure.