Predicting Bid Verification in Spectrum Auctions: A Data-Driven Approach

Abstract

Spectrum auctions are very important for the strategic allocation of frequency bands in the telecommunications industry, ensuring efficient and fair access to this valuable resource. However, the complexity of auction environments—characterized by vast state spaces and multidimensional bid attributes—renders manual bid verification infeasible. This study introduces an innovative, data-driven approach by utilizing machine learning models, including k-nearest neighbors, support vector machines, decision trees, and stochastic gradient descent classifiers, to automate the verification process. Through hyperparameter tuning and rigorous k-fold cross-validation, the decision tree model emerged as the most effective, achieving an F1-score of 96% and a G-Mean of 97%. These results demonstrate the practical viability of AI-enhanced verification systems in spectrum auctions and suggest broader applicability across various high-stakes auction platforms where real-time, reliable validation is essential.