| With the development of economy and the continuous change of people’s life needs,the scorecard model developed in the financial field has brought convenience to many industries.For example,banks can use the scorecard model to avoid some credit risks and help enterprises in a healthy state of development.In the field of finance,scholars have widely combined machine learning methods and scorecard models with the advantage of clearly labeled credit data,and continuously enriched the types and applicable scope of models.At present,most of the research background of scorecard model focuses on the finance and insurance industry.The well-known field of vehicles,especially the field of Internet of Vehicles,is also rapidly developing.It is essential to study the factors that affect the safety of vehicle driving,such as emotion,age,gender,road conditions,weather,driving age,etc.In addition,the vehicle speed is studied to obtain the relationship between side friction factor and speed.Then,the driving feature can be classified,which lays a foundation for the depth mining of data and the construction of user portraits.It is of great practical significance to migrate the financial scorecard model to the field of Internet of Vehicles and establish the driving behavior safety scorecard model,so as to evaluate driving safety.In this thesis,the Internet of Vehicles data in a city are selected to establish driving behavior scorecard.Through data preprocessing,data preparation for the construction of scorecard is completed.This thesis first establishes the driving safety scorecard based on Logistic regression.In the scenario of Logistic regression modeling,the judgment criterion of predicting dangerous driving behavior is given.Consider the importance of feature selection for scorecard modeling.Then the MIV feature selection algorithm based on BP neural network in this thesis,establish XGBoost driving safety scorecard.And the judgment criterion in this modeling scenario is given. |
PDF Full Text Request