The rapid growth of urban populations has intensified parking demand, leading to poor management, congestion, wasted time, and increased environmental impacts. Traditional parking systems are often unable to provide real-time, dynamic recommendations, resulting in overcrowding and higher carbon emissions. To address these inefficiencies, this study proposes a real-time parking spot recommendation system based on Artificial Intelligence (AI) and Big Data Analytics. The system gathers data from multiple sources, including parking sensors, IoT devices, cameras, and human inputs, which undergo preprocessing cleaning and normalization—to remove noise and ensure authenticity. Features are then extracted using the Fourier Transform (FT), capturing periodicity and trends in parking behavior, which is crucial for predicting busy rates more accurately. To further enhance predictions, a hybrid model named Flexible Tabu Search-based Customized Naive Bayes (FT-CNB) is introduced, where Tabu Search Optimization leverages its global search capacity to refine Naive Bayes predictions, improving accuracy while reducing computation time. This integration enables the system to provide faster and more precise real-time recommendations for available parking spots. The model’s performance was evaluated using multiple metrics, achieving an accuracy of 95.91%, precision of 96.53%, recall of 97.21%, and an F1-score of 93.07%, demonstrating that the FT-CNB model is robust and reliable. Overall, the proposed approach significantly improves real-time parking management, reducing congestion, minimizing time wasted searching for parking, and mitigating negative environmental externalities.
Shetra et al. (Thu,) studied this question.