Wireless traffic prediction is vital for network planning and management, enabling real-time decisions and accurate short- and long-term forecasts. Effective, cost-efficient prediction techniques are essential for improving cellular networks and ensuring high-quality service through optimal resource allocation. Future network requirements can be predicted by detecting regular traffic patterns, making proactive management strategies like load balancing and coordinated beam management possible. In this paper, eight machine learning models were developed and evaluated for cellular network traffic prediction: Seasonal-Autoregressive-Integrated-Moving-Average (SARIMA), Facebook-Prophet (FB Prophet), Adaptive-Boosting (AdaBoost), Extreme-Gradient-Boosting (XGBoost), Long-Short-Term Memory (LSTM), Convolutional-Neural-Networks (CNN), hybrid CNN-LSTM, and an ensemble model combining CNN and LSTM. These models were compared based on performance metrics and computational time. According to the results, deep learning models despite their longer computation times, demonstrated better prediction accuracy than other learning models, such as CNN, LSTM, hybrid CNN-LSTM, and ensemble CNN + LSTM. Models for machine learning, such as XGBoost and AdaBoost, offered a fair compromise between computational time and prediction accuracy. Conversely, the SARIMA statistical model provided the least accurate prediction.