Research On Non-systematic Risk Prediction Of Real Estate Trust Project Of Company X

Non-systematic risk assessment of real estate trust projects has important policy and practical significance.Existing literature mostly uses expert scoring method or tomographic analysis method to judge non-systemic risks of projects,which is subjective and has low review efficiency.This paper explores and compares the predictive ability of machine learning models for non-systemic risks of real estate trust projects,with a view to obtaining project non-systemic risk prediction models,eliminating subjective factors of risk assessment mailboxes,and improving project risk assessment efficiency.This paper collects the actual real estate trust project data of Company X,extracts the characteristic variables,obtains the prediction results and compares them with the actual risk assessment results.It is found that the Logistic Ridge regression model is the optimal project non-systematic risk assessment model.The structure of this article is as follows: First,the importance of risk assessment of real estate trust projects is expounded from the two dimensions of actual and policy requirements,and related literature is combed to introduce the research significance of the article.Secondly,it briefly introduces the development history,development status and project characteristics of real estate trust projects.Afterwards,the real estate trust project was identified for risk,and the controllable non-systemic risk was selected as the research object,and the non-systemic risk composition was sorted out based on the WBS-RBS method;a combination of literature research and case analysis was used to identify various types of non-systematic projects Characteristic variables of sexual risk.Based on the selected feature variables,the project data is collected and preprocessed to obtain sample data.The LR model,BP neural network and support vector machine in machine learning are selected respectively to construct the non-systematic risk binary classification prediction model of the project,and the prediction accuracy rate,F value,AUC and comprehensive cost pair prediction model are selected,and the Logistic Ridge is found.Regression model is the current best project non-systemic risk prediction model,according to the Logistic Ridge regression model results to put forward risk control opinions.