Time Series Prediction System For Cellular Network Traffic Data Using Long Short-Term Memory, Machine Learning, and Statistics Model

The fast growth of cellular network traffic caused by data-heavy apps and IoT devices has made it very hard for network operators to make sure that resources are used efficiently and the network stays stable. For managing bandwidth allocation, avoiding congestion, and lowering operating costs, it's important to be able to accurately predict traffic. This study assesses the efficacy of statistical, machine learning, and deep learning models for predicting cellular network traffic. We used and compared models like ARIMA, SARIMA, Random Forest, Gradient Boosting, and Long Short-Term Memory (LSTM) based on important metrics like Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), and R-squared (R²). The results show that LSTM did better than all the other models by being the most accurate and reliable at finding temporal dependencies and nonlinear traffic patterns. Machine learning models such as Gradient Boosting and Random Forest exhibited competitive performance, whereas statistical models revealed deficiencies in managing intricate data structures. A user-friendly dashboard was made to combine data preprocessing, model selection, and result visualization, which helped network operators make smart choices. Future work includes making the system able to make real-time predictions, making it work with larger datasets and 5G networks, and looking into hybrid models to improve performance even more. This study offers a thorough framework for traffic prediction and functions as a practical instrument for enhancing network management in contemporary cellular settings.