Conference paper delivered at <a href="https://ieeexplore.ieee.org/xpl/conhome/1000225/all-proceedings">IEEE Canadian Conference on Electrical and Computer Engineering, Vancouver BC, (26-29 May, 2025).</a>
IT professional's first job after graduation is a significant milestone, marking the first step to financial independence. To explore the factors influencing compensation for IT graduates, a binary classification model using an ensemble machine learning approach that integrates Random Forest (RF), Neural Networks (NN), and LightGBM was developed. A multi-level strategy was adopted, beginning with training the data using RF and subsequently feeding the output leaf indices and the feature set into NN, culminating with LightGBM functioning as a meta-classifier. A cross-validation approach was employed to assess the model's accuracy rigorously. The model achieved an 88% accuracy rate and an F1 score exceeding 80% across all categories. Utilizing SHAP analysis, key features per model were extracted and analyzed. Notable features highlighted by the two models are the mother's educational level, IT experience, degree concentration, study frequency, accommodation, siblings and grades. Features such as holding a degree in cybersecurity and residing on dorms emerged as significant predictors of higher starting salaries. The model offers valuable insights for students, enabling them to enhance their qualifications and improve their compensation prospects after graduation.
