Breakthroughs in artificial intelligence, machine learning anddeep learning with great recent achievements have been strongly impacting the mobile communication industry. The traditional method ofmobile network management and operation has been shifting toward automation and proactive monitoring. Forecasting radio mobile networks’behaviors in coexisting multi-layer technology mobile networks for improving resource allocation efficiency and optimizing network’s key performance indicators becomes a critical task. This paper proposes usingLong-Short Term Memory (LSTM) network for predicting radio networks’ behaviors based on real statistical data from operational supportsystems. The promising results show the potential of our approach. Theprediction results can help mobile network operators leverage networks’operating performance.