Statistical Analysis Of Users' Drving Behavior Based On Internet Of Vehicles

In 2005,the concept of the Internet of Things carried out and has been widely used.The Internet of Vehicles technology is one of the most important application types of the Internet of Things.Vehicle networking refers to a network that intelligently identifies,locates,tracks,monitors,and manages all vehicles connected to the Internet through information sensing equipment in accordance with agreed protocols for information exchange and communication.A large amount of data analysis shows that users' daily driving behaviors and habits can affect travel safety.By analyzing the vehicle observation data obtained by the Internet of Vehicles technology,the user's driving behavior and travel habit characteristics can be analyzed.These characteristics have important practical significance for automobile companies and users.The thesis mainly includes two parts.In the first part,we tap driving indicators and build user portraits.Firstly,based on the seven valid indicators in the original data,two categories of secondary indicators were constructed: driving style and travel habits.Among them,four and eleven three-level refinement indicators were excavated under each of the second-level indicators.Secondly,the 11 three-level refinement indicators of travel habits are directly programmed and calculated.Cluster analysis,interval estimation and difference method were used to obtain four three-level indicators under the driving style secondary indicators.Finally,each indicator definition is encapsulated into a function,which is substituted into 80 users' data,and 80 users' driving behavior portraits are obtained.In another part,we judge driving behavior based on user portraits.Firstly,the principal component analysis was carried out for 11 indicators of 80 users travel habits,and 11 indicators are reduced to 4 principal component factors: “use frequency” factor,“peak-fatigue” factor,“fatigue driving” factor and “early-night peak” factor which were used to determine the type of travel habits of the users.Secondly,cluster analysis was conducted on four indicators of driving style,and 80 users were grouped into 4 categories: "safe","dangerous","robust" and "impatient",to judge the driving style of users.Finally,the scoring card model of Logistic regression was established for the 15 three-level indicators of driving style and travel habits,and users' driving behaviors were evaluated by giving their scores.In this way,effective information can be provided for automobile production companies.By combining different types of users' driving characteristics and rating levels,automobile companies can carried out product optimization and provide personalized service for their customers.At the same time,automobile users can seek personal product recommendations(such as insurance types,etc.)from automobile companies based on their driving habits and travel characteristics.