Hybrid PSO-Adaptive Boosting Regression for Employee Salary Prediction and Recommendation

M. Amran Hakim Siregar (Universitas Mitra Bangsa)
Bachtiar Ramadhan (Universitas Logistik dan Bisnis International)
Syafrial Fachri Pane (Universitas Logistik dan Bisnis International)

Nuansa Informatika
Vol. 20 No. 2 (2026): Nuansa Informatika 20.2 July 2026

Recommending appropriate employee salaries is important for supporting employee performance and data-driven managerial decisions. This study develops a hybrid machine learning model to recommend employee salaries and identify influential factors affecting monthly income. The dataset was obtained from Kaggle and consisted of 1,029 employee records with 34 variables covering company, personal, and demographic characteristics. Data preprocessing included categorical encoding, missing-value handling, duplicate checking, and outlier removal using the Interquartile Range method. The proposed approach combines Particle Swarm Optimization for variable optimization with an AdaBoost Regressor selected through TPOT Regression. Model performance was evaluated using R-Square and Mean Absolute Percentage Error. The PSO-AdaBoost Regressor achieved an R-Square value of 0.88 and a MAPE value of 0.22. Feature importance analysis identified Job Level as the most influential feature, with a score of 0.97156. The results were implemented in a Django-based web application