Salary Prediction with Ensemble Regressor Model

Salary prediction is very important for companies and job seekers to know the estimated salary for a specific job. This study will focus on predicting the estimated salary for a professional employee by using ensemble regressor models to get the exact salary value. We will use RFR and XGBR as a single model and combine them in a hybrid model by using the Stacking Regressor, Bagging Regressor, and Voting Regressor. The data used in this research comes from Kaggle, which are Glassdoor and DS Salaries dataset with more than ten features. After preprocessing the data, training, and evaluating using MAE, MAPE, R2, and WMAPE. The result shows that the hybrid ensemble regressor achieves the best result, which Voting Regressor achieves the lowest MAE 15,376, MAPE 15.88%, and WMAPE 15.67%, also the best R2 score of 0.38 in Glassdoor dataset, but in DS Salaries dataset, Bagging Regressor achieves the lowest MAE 38,361, MAPE 35.55%, and WMAPE 26.13%, also the best R2 score of 0.44. The proposed model improvement achieves a more accurate and reliable model for salary predictions. However, the model’s overall performance was not as strong as anticipated, with moderate gains in error rates and R2 scores.