Analysis of Passenger Satisfaction Evaluation Metrics Based on Machine Learning

The research in this paper is based on a dataset on airline passenger satisfaction downloaded from the Kaggle data platform, which contains 26 characteristics of passengers, totaling nearly 130,000 records. Using this data set, eight machine learning algorithms are used, such as GBDT, Decision tree, AdaBoost, XGBoost, Random forest, Extremerandom tree, CatBoost, HistGBDT. By analyzing the evaluation indexes of these 8 machine learning algorithms, it is concluded that HistGBDT algorithm has higher values in precision and accuracy. This result page verifies that HistGBDT model can improve the predictive ability of passenger satisfaction. Increase the generalization ability of their models. It also shows that the model is superior in predicting airline satisfaction. Through this result analysis, we can help airlines to better provide passengers with better quality services and meet their different needs, so as to provide airlines with more personalized services and decisions, and enhance their competitive advantages.