Explainable Machine Learning for Detecting Fraudulent Online Job Postings in Ghana: A SHAP-Based Approach

Abstract Employment fraud through online job postings is an increasing concern in Ghana, and traditional detection methods are often inefficient and lack scalability. This study applies machine learning and Explainable Artificial Intelligence (XAI) techniques to detect fraudulent job advertisements while enhancing model interpretability. Using a dataset of 499 job postings from Jobweb Ghana, five classification models Multinomial Naive Bayes, Logistic Regression, Random Forest, K-Nearest Neighbors, and Decision Tree were developed and evaluated using accuracy, precision, recall, F1-score, and ROC--AUC metrics. SHapley Additive exPlanations (SHAP) were employed to interpret model predictions. The results show that Multinomial Na\"ive Bayes and Logistic Regression achieved the best performance. SHAP analysis identified educational requirements, application deadline disclosure, and job categorization as key indicators of employment fraud. The study demonstrates that explainable machine learning models can effectively detect employment fraud while providing transparent and interpretable insights.