This study proposes a neural network ensemble method for accurate path loss prediction in cellular networks. Path loss, which decreases the strength of a radio signal between a base station and mobile station, is crucial for base-station positioning. Traditional methods rely on time-consuming field tests, so the authors introduce a machine learning (ML)-based approach using neural network ensemble learning to enhance prediction accuracy and performance. The method involves selecting top-ranked neural networks based on hyperparameter optimization results and integrating their predictions to improve overall performance. The proposed method was tested on a public dataset and outperformed various ML-based methods, including SVM, k-NN, random forest, and artificial neural networks. The results showed that the ensemble model achieved higher accuracy and lower error rates compared to individual models. The study also details the dataset preparation, feature scaling, and hyperparameter optimization processes, demonstrating the effectiveness of the ensemble approach in predicting path loss. The method's performance was evaluated using metrics such as MSE, RMSE, MAE, MAPE, MSLE, RMSLE, and R², with the ensemble model showing the best results across all metrics. The proposed method provides a robust and accurate solution for path loss prediction, with potential applications in improving the efficiency and reliability of cellular network operations.This study proposes a neural network ensemble method for accurate path loss prediction in cellular networks. Path loss, which decreases the strength of a radio signal between a base station and mobile station, is crucial for base-station positioning. Traditional methods rely on time-consuming field tests, so the authors introduce a machine learning (ML)-based approach using neural network ensemble learning to enhance prediction accuracy and performance. The method involves selecting top-ranked neural networks based on hyperparameter optimization results and integrating their predictions to improve overall performance. The proposed method was tested on a public dataset and outperformed various ML-based methods, including SVM, k-NN, random forest, and artificial neural networks. The results showed that the ensemble model achieved higher accuracy and lower error rates compared to individual models. The study also details the dataset preparation, feature scaling, and hyperparameter optimization processes, demonstrating the effectiveness of the ensemble approach in predicting path loss. The method's performance was evaluated using metrics such as MSE, RMSE, MAE, MAPE, MSLE, RMSLE, and R², with the ensemble model showing the best results across all metrics. The proposed method provides a robust and accurate solution for path loss prediction, with potential applications in improving the efficiency and reliability of cellular network operations.