FAIR AND TRANSPARENT STUDENT PLACEMENT PREDICTIONS: A MACHINE LEARNING AND EXPLAINABLE AI (XAI) APPROACH

Abstract

Career path prediction plays a vital role in guiding students towards successful careers by offering personalized recommendations based on academic performance, skills, interests, and market trends. Early career path identification allows students to develop relevant skills and gain necessary experience, increasing their competitiveness in the job market. This study aims to ensure fair and transparent predictions by leveraging Machine Learning (ML) and Explainable AI (XAI) techniques on student career dataset. Initially, ML algorithms were applied to predict placement status, followed by an assessment of the model for bias using XAI. Upon detecting bias, mitigation strategies were implemented to enhance fairness. The use of XAI techniques improved model transparency and trustworthiness, allowing stakeholders to understand and trust the decision-making process. The methodology involved identifying and addressing dataset imbalances that could skew predictions. By applying oversampling techniques, the dataset was balanced, leading to significant improvements in model performance. The initial model showed poor performance metrics due to data imbalance, but after oversampling, the F1 score improved. Further application of XAI techniques, such as Local Interpretable Model-agnostic Explanations (LIME) and SHapely Additive exPlanation (SHAP), provided deeper insights into the model's decision-making process. This analysis highlighted specific features that, when oversampled, further enhanced the F1 score. The study emphasizes the importance of using XAI to not only improve mod el performance but also provide a trustworthy framework for stakeholders. By evaluating models using metrics like recall, precision, accuracy, and F1 score, the study demonstrated that integrating fairness and transparency into predictive models is achievable and beneficial for ensuring equitable student placement outcomes.