Can Artificial Intelligence Help Tourism Enterprises Identify Their Business Performance?

Due to increasingly fierce competition in the tourism market,traditional tourism enterprises are facing huge,immediate challenges.They must comparatively analyze and forecast their own business performances,break through development bottlenecks,and seize opportunities for long-term development.Researchers have proposed many algorithmic models for this kind of problem and are continuously enhancing the classification evaluation model of business performance,using traditional mathematical statistics models and artificial intelligence algorithms.The traditional mathematical statistics model is relatively mature in terms of the research methods and forms a more complete research system.However,due to strong assumptions and limited data,the accuracy and reliability of the traditional econometric models are reduced when models are applied to practical problems.Artificial intelligence algorithms can learn from a large number of different types and dimensions of data,and quickly adapt to complex environments.These kinds of models have become very attractive to researchers,and are widely used in sales forecasting,stock analysis,customer service,et cetera.Four common and widely used models were adopted for this study:Logistic Regression Model(LR),Support Vector Machine(SVM),Random Forest(RF)and Gradient Boost Descent Tree(GBDT).This paper selects 25 listed tourism companies in China for analysis and compares the performance of each classification evaluation model as determined by AUC values,accuracy,precision,recall,and Type I/II errors.Finally,in order to reduce the bias of each independent classification evaluation model,the author used Stacking integration technology to combine the above four classification evaluation models and improve their accuracy,thus making the conclusion more reliable.This paper hopes to understand the differences between different models by studying different classification evaluation models,and provide reference for exploring the best performance of classification evaluation models by tourism enterprises in the future.This can also help stakeholders better understand the status of the enterprise and development of the industry market,and conduct more rational and effective decision-making.