To accurately predict the flight transit key nodes time such as departure and takeoff,and improve the operational efficiency of busy airports,a multi-time prediction model of flight transit key nodes time based on gradient boosting decision tree (GBDT) is proposed in this paper. Firstly,the flight information data items are classified according to the generation time. Secondly,based on the GBDT algorithm and Spark platform,the prediction models of flight transit key nodes time at different transit times are constructed respectively. Finally,flight data is obtained and pro-cessed in real-time computing manner,enabling dynamic prediction of flight departure time and take-off time at multiple times. The experimental results show that the proposed model has good predictive performance and has the bestpredictiveperformancecomparedtootheralgorithms,withapredictionaccuracyof95.6％ within±15minutes.
维普
万方数据
