Globalization contributes to the high expansion of the transport industry. This is particularly evident in air traffic on the smaller airports that lack the costly specialized systems that accommodate nowadays needs of increased travel. One aspect, that needs to be addressed, is a prediction of Time of Arrival and any potential delays and obstacles that may influence the trajectory of the travel. Every minute of the wrong prediction may cause unnecessary expenses and management problems. Unfortunately, despite the availability of aircraft movement information to air traffic control or airlines, it often is not communicated further, leading to unnecessary costs in management and contingency planning. Hence, we propose an algorithm for the aircraft trajectory prediction as well as Time of Arrival based on Social-LSTM (Long short-term memory) approach introduced for the pedestrian traffic prediction~\cite{soomer_scheduling_2008}. In this approach, the motion sequence and arrival time are predicted not only based on historical trajectories, but also considering the trajectories of neighbouring aircraft. Social-LSTM approach extends on LSTM network with an additionally embedded hidden-state tensor, which provide the coordinates of neighbouring aircraft and their trajectories. Our experiment suggests that this approach decreases the dimensionality of data needed to predict the detailed trajectory (weather data and unexpected events data) and gives more accurate prediction compared to normal LSTM approaches, especially for unusual weather conditions or events. Approach was carried with many variants like basic LSTM, GRU (Gated Recurrent Unit) as well as weather data embedded as 4D weather cubes variant.