Decision Tree Model And Its Application In The Prediction Of Emergency Stop Of Capital Inflow

Artificial intelligence has been widely discussed in the whole society in the past decade.Machine learning,as a sub field of artificial intelligence,has also been more popular at this opportunity.It has been applied to practice by experts in many fields,such as medicine,and achieved excellent results.Among many machine learning algorithms,XGBoost algorithm has been paid special attention in recent years.Since Chen Tianqi proposed XGBoost algorithm in 2012,XGBoost has become the most popular machine learning algorithm with its superior performance in many competitions in kaggle and Ali Tianchi.Theoretically,XGBoost algorithm has better controllability and can effectively prevent over fitting.From the perspective of data,the model based on decision tree can reduce the influence of noise in data to a certain extent.In terms of implementation,XGBoost algorithm is more accurate and efficient.Although XGBoost algorithm has shown great advantages in various fields of practical problems,machine learning is rarely used in the field of economics.On the one hand,the access to complete international economic data is limited.On the other hand,there is a certain correlation among the economic indicators themselves,which increases obstacles for the application of machine learning in this field.But the XGBoost algorithm based on decision tree can eliminate the disturbance of missing data and noise data to a certain extent,and the correlation characteristics have no obvious impact on the XGBoost algorithm.Therefore,XGBoost algorithm is expected to have good performance in the capital inflow emergency stop prediction problem.Therefore,the main purpose of this article is to find the key features with a large number of economic indicators,and to determine and predict the emergency capital suspension.First,according to relevant theories and foreign research results,determine the characteristics that affect the emergency stop of capital inflows,including domestic,international and external factors,and collect data through existing data disclosures and conduct data processing and analysis.Taking into account the advantages of XGBoost in missing data processing,generalization ability,controllability,and scope of application,this paper chooses the XGBoost algorithm for modeling,and adjusts the parameters of the model through grid search to predict the emergency stop of capital inflows.Finally,Comparing the prediction results with the logistic regression model,it can be found that XGBoost is more efficient and has a better prediction effect.The result of this research is an attempt to study the problem about prediction of sudden stop of capital inflow,which provides the possibility for accurately predicting and improving the ability of countries in the world to deal with economic problems.At the same time,it also shows that machine learning method is a meaningful practice when studying the related economic problems.