Salary prediction using machine learning

This research explores how individual and demographic factors, such as professional experience, location, and education, influence salary trends and aims to uncover insights into their relative importance. Accurately predicting salaries helps understand trends in the workforce and assists career planning. Machine learning offers powerful tools for analyzing complex relationships between factors such as basic demographics, and wages. A synthetic dataset of 1,000 samples with 6 features is used. The performance of three neural network models is evaluated through test loss values, with the model incorporating a single hidden layer achieving the lowest loss of 0.1415. The model with no hidden layer and 2 hidden layers recorded test losses of 0.1435 and 0.145 respectively. Among the features, high school education, PhD, and years of experience demonstrate high contributions to salary predictions, as measured by permutation importance. This study demonstrates the efficiency of neural networks in predicting salaries, even with a synthetic dataset, showcasing their ability to generalize well and outperform manual prediction methods. Beyond prediction accuracy, interpretability techniques like permutation importance provide valuable insights into determinants of salaries for job seekers and employers. This study reinforces the effectiveness of machine learning in salary prediction.