基于IGA-Optuna-LightGBM的民航潜在旅客预测

Potential passenger forecasting for civil aviation based on IGA-Optuna-LightGBM

为了进一步提升民航潜在有价值旅客的预测准确度,提出一种基于LightGBM的民航潜在旅客预测模型。首先,通过改进遗传算法的选择算子和交叉变异概率,改善标准遗传算法易于陷入局部最优和收敛速度慢的问题,并使用改进遗传算法(IGA)进行特征选择,找到最优特征变量;其次,对LightGBM模型进行训练,使用Optuna框架优化超参数,得到最终的旅客预测模型;最后,通过LightGBM模型对民航旅客进行类型预测,进而找到具有潜在价值的旅客。实验结果表明,基于IGA-Optuna-LightGBM模型的预测准确度达到0.962,AUC值达到0.991,预测性能优于其他模型。

In order to further improve the accuracy of forecasting potential valuable passengers in civil aviation, a forecasting model of civil aviation potential passengers based on LightGBM is proposed. Firstly, by improving the selection operator and cross mutation probability of genetic algorithm, the standard genetic algorithm is easy to fall into local optimum and slow convergence speed, and the improved genetic algorithm(IGA) is used for feature selection to find the optimal feature variable. Secondly, the LightGBM model is trained, and the Optuna framework is used to optimize the superparameters to obtain the final model. Finally, using the test set, the type of civil aviation passengers is predicted by the LightGBM model, and then the passengers with potential value are found. The experimental results show that the prediction accuracy of IGA-Optuna-LightGBM model is 0.962, AUC value reached 0.991, and the prediction performance is superior to other models.