WO2020015526A1

WO2020015526A1 - Driving behavior scoring method and device

Info

Publication number: WO2020015526A1
Application number: PCT/CN2019/094500
Authority: WO
Inventors: 梅鵾; 陆璐; 谢畅; 钱浩然; 孙谷飞
Original assignee: 众安信息技术服务有限公司
Priority date: 2018-07-19
Filing date: 2019-07-03
Publication date: 2020-01-23
Also published as: CN109118055A; JP2020530578A; CN109118055B; SG11202002191SA

Abstract

A driving behavior scoring method and device. The method comprises: obtaining driving behavior data, driving environment data, and service insurance policy data of a target driver (110); determining target feature data on the basis of the driving behavior data, the driving environment data, and the service insurance policy data of the target driver (120); and inputting the target feature data into a driving behavior scoring model to obtain a driving behavior score of the target driver (130). The driving behavior scoring method and device can improve the accuracy of the scoring result and further improves the accuracy of vehicle insurance pricing.

Description

Method and device for driving behavior scoring

Technical field

The invention relates to the technical field of vehicle driving behavior analysis, in particular to a driving behavior scoring method and device.

Background of the invention

In recent years, with the increasing number of motor vehicles and the rapid development of the Internet of Vehicles, there are more and more auto insurance products that are priced based on driver driving behavior combined with Internet of Vehicles technology. By scoring the driver's driving behavior and then accurately determining the pricing of auto insurance products, the driver's safe driving consumption outlook and driver's safe driving awareness can be improved, thereby reducing the social traffic accident rate. The data collected by the current driving behavior scoring methods are relatively single, and the scoring results are not accurate enough, which makes it difficult to reasonably price auto insurance products.

Summary of the invention

In order to solve the above technical problems, embodiments of the present invention provide a driving behavior scoring method and device, which can improve the accuracy of the scoring results, thereby improving the accuracy of auto insurance pricing.

The specific technical solution of the embodiment of the present invention is as follows:

In a first aspect, an embodiment of the present invention provides a driving behavior scoring method, including: obtaining driving behavior data, driving environment data, and business insurance policy data of a target driver; based on the driving behavior data, driving environment data, and business of the target driver The policy data determines the target characteristic data; the target characteristic data is input into the driving behavior scoring model to obtain the driving behavior score of the target driver.

In some embodiments of the present invention, obtaining the driving behavior data, driving environment data, and business warranty data of the target driver includes: obtaining the driving behavior data of the target driver through a mobile software development kit module; and obtaining the vehicle depth camera module Target driver's driving environment data; business policy data of the target driver is obtained through a preset interface.

In some embodiments of the present invention, determining the target characteristic data based on the driving behavior data, driving environment data, and business insurance data of the target driver includes: extracting the first characteristic of the target driver from the driving behavior data of the target driver. The first feature includes at least one of the average driving mileage, the average driving speed, the number of rapid accelerations, the number of rapid decelerations, the number of sharp turns, and whether or not the driver is fatigued; the second Feature, the second feature includes at least one of driving conditions and lane changing times; third and fourth features of the target driver are extracted from target driver's business insurance data, the third feature includes basic personal information, The four features include at least one of policy purchase record information and claim history information; the first feature, the second feature, the third feature, and the fourth feature are combined to determine target feature data.

In some embodiments of the present invention, the method of the first aspect further includes: training the machine learning model using sample data to obtain a driving behavior scoring model.

In some embodiments of the present invention, training machine learning models using sample data to obtain driving behavior scoring models includes: obtaining sample data, the sample data including driving behavior data, driving environment data, and business insurance data of sample drivers; based on samples The driver's driving behavior data, driving environment data, and business policy data determine the sample characteristic data and the sample driver's driving behavior score; train the machine learning model based on the sample characteristic data and the sample driver's driving behavior score to obtain the driving behavior score model.

In some embodiments of the present invention, determining the sample characteristic data and the driving behavior score of the sample driver based on the driving behavior data, driving environment data, and business policy data of the sample driver includes: based on the driving behavior data of the sample driver, driving The environmental data and business insurance data determine the sample characteristic data; based on the business insurance data of the sample driver, the driving behavior score of the sample driver is determined.

In some embodiments of the present invention, determining the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver includes: extracting a fifth characteristic of the sample driver from the driving behavior data of the sample driver Extract the sixth characteristics of the sample driver from the driving environment data of the sample driver; extract the seventh and eighth characteristics of the sample driver from the business insurance data of the sample driver; merge the fifth and sixth characteristics Feature, seventh feature, and eighth feature, determining sample feature data; wherein determining the driving behavior score of the sample driver based on the business insurance data of the sample driver includes determining driving of the sample driver based on the seventh feature and the eighth feature Behavioral scoring.

In some embodiments of the present invention, determining the driving behavior score of the sample driver based on the seventh feature and the eighth feature includes: calculating the claim rate of the sample driver according to the seventh feature and the eighth feature; and according to a preset The mapping relationship table determines the driving behavior score of the sample driver having a mapping relationship with the claim rate. The higher the claim rate, the lower the driving behavior score of the sample driver.

In some embodiments of the present invention, the machine learning model is a boosted tree model or a random forest model.

In a second aspect, an embodiment of the present invention provides a driving behavior scoring device, including: an acquisition module for acquiring driving behavior data, driving environment data, and business warranty data of a target driver; and a determination module for using the target driver based on The driving behavior data, driving environment data, and business insurance data determine the target characteristic data; the scoring module is used to input the target characteristic data into the driving behavior scoring model to obtain the driving behavior score of the target driver.

In some embodiments of the present invention, the obtaining module is configured to: obtain the driving behavior data of the target driver through a mobile software development kit module; obtain the driving environment data of the target driver through a vehicle depth camera module; and obtain the target through a preset interface Driver's business policy data.

In some embodiments of the present invention, the determining module is configured to extract a first characteristic of the target driver from the driving behavior data of the target driver, and the first characteristic includes an average driving mileage, an average driving speed, a number of rapid accelerations, and an emergency At least one of the number of decelerations, the number of sharp turns, and whether fatigue driving; the second characteristic of the target driver is extracted from the driving environment data of the target driver, and the second characteristic includes at least one of the driving state and the number of lane changes; The third and fourth characteristics of the target driver are extracted from the target driver's business policy data. The third characteristic includes basic personal information, and the fourth characteristic includes at least one of policy purchase record information and claim history information; A feature, a second feature, a third feature, and a fourth feature determine target feature data.

In some embodiments of the present invention, the apparatus of the second aspect further includes: a training module for training a machine learning model using sample data to obtain a driving behavior score model.

In some embodiments of the present invention, the training module is configured to: obtain sample data, the sample data including driving behavior data, driving environment data, and business warranty data of the sample driver; based on the driving behavior data, driving environment data, and The business policy data determines the sample characteristic data and the driving behavior score of the sample driver; the machine learning model is trained according to the sample characteristic data and the driving behavior score of the sample driver to obtain the driving behavior score model.

In some embodiments of the present invention, the training module is configured to determine the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver; determine the driving behavior of the sample driver based on the business insurance data of the sample driver score.

In some embodiments of the present invention, the training module is configured to: extract the fifth characteristic of the sample driver from the driving behavior data of the sample driver; extract the sixth characteristic of the sample driver from the driving environment data of the sample driver; Features; the seventh and eighth features of the sample driver are extracted from the sample driver's business insurance data; the fifth, sixth, seventh, and eighth features are combined to determine the sample feature data; based on the seventh The feature and the eighth feature determine the driving behavior score of the sample driver.

In some embodiments of the present invention, the training module is configured to: calculate the claim rate of the sample driver according to the seventh feature and the eighth feature; and determine a sample drive having a mapping relationship with the claim rate according to a preset mapping relationship table Driver ’s driving behavior score, in which the higher the claim rate, the lower the driving behavior score of the sample driver.

According to a third aspect, an embodiment of the present invention provides a computer device including: one or more processors; a memory configured to store one or more programs; and when one or more programs are executed by one or more processors, One or more processors are caused to implement the driving behavior scoring method according to the first aspect.

According to a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the driving behavior scoring method according to the first aspect is implemented.

Embodiments of the present invention provide a driving behavior scoring method and device. Starting from multiple dimensions of a target driver's driving behavior data, driving environment data, and business policy data, the target characteristic data of the target driver is extracted and used. The behavior scoring model predicts the driving behavior score of the target driver, which can improve the accuracy of the scoring results, thereby improving the accuracy of auto insurance pricing.

Brief description of the drawings

In order to explain the technical solutions in the embodiments of the present invention more clearly, the drawings used in the description of the embodiments are briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained according to these drawings without paying creative labor.

FIG. 1 is a schematic flowchart of a driving behavior scoring method according to an embodiment of the present invention.

FIG. 2 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention.

FIG. 3 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention.

FIG. 4 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention.

FIG. 5 is a schematic structural diagram of a driving behavior scoring device according to an embodiment of the present invention.

FIG. 6 is a block diagram of a computer device according to an embodiment of the present invention.

Mode of Carrying Out the Invention

In order to make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are merely Some embodiments of the present invention are not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

At present, driving behavior score modeling is mainly to obtain user driving behavior data through docking sensors or vehicles, and to obtain data on whether the vehicle has a traffic accident during the driving behavior observation cycle and the user's application for insurance claims from insurance companies. The classification method is used to model the data to predict the risk probability, and then, the mapping relationship between the driving behavior score and the risk probability is established. However, this driving behavior scoring modeling method mainly has the following problems: First, the collected data sources are relatively single and only contain driving behavior-related information, which leads to inaccurate scoring results; second, the continuity of the vehicle owner cannot be guaranteed. Multiple users can be registered under a car, causing the driving records of the same car to not belong to the same user, which interferes with the analysis; the third is to set the labeling of driving behavior only by the probability of risk, and the lack of measurement of the degree of risk, resulting in The model is not refined enough.

FIG. 1 is a schematic flowchart of a driving behavior scoring method according to an embodiment of the present invention. Referring to FIG. 1, the method includes the following content.

110: Obtain driving behavior data, driving environment data, and business policy data of the target driver.

Specifically, the execution subject of the driving behavior score may be a server, the server may be a single server, or a server group composed of multiple servers, and multiple servers in the server group may communicate with each other.

The server can communicate with the vehicle-mounted system through the network. The vehicle-mounted system can obtain the driving behavior data, driving environment data, and business warranty data of the target driver, and then transmit the data to the server.

Driving behavior data can include information such as speed, acceleration, and steering; driving environment data can include information such as road conditions and lane detection; business policy data can include basic information about target drivers, policy purchase information, and claim information.

120: Determine target characteristic data based on the driving behavior data, driving environment data, and business policy data of the target driver.

Specifically, features can be extracted from driving behavior data, driving environment data, and business policy data and feature combination can be performed to obtain target feature data. The target feature data may be a vector.

130: Input the target characteristic data into the driving behavior scoring model to obtain the driving behavior score of the target driver.

Specifically, the driving behavior scoring model may be obtained by training a machine learning model. In this embodiment, the driving behavior scoring model stored offline can be called and the target characteristic data is input to the driving behavior scoring model. The driving behavior scoring model can predict the driving behavior score of the target driver and the driving behavior of the target driver The score is output for invocation by other devices that interact with the server (for example, a server at an insurance company).

An embodiment of the present invention provides a driving behavior scoring method. The driving characteristic data of the target driver is extracted from multiple dimensions of the driving behavior data, driving environment data, and business policy data of the target driver, and the driving behavior score is used. The model predicts the driving behavior score of the target driver, which can improve the accuracy of the scoring results, and then improve the accuracy of auto insurance pricing.

FIG. 2 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention. The embodiment shown in FIG. 2 is an extension of the embodiment shown in FIG. 1. In order to avoid repetition, the differences are highlighted here. Place. As shown in FIG. 2, 110 in FIG. 1 includes the following content.

111: Obtain driving behavior data of a target driver through a Mobile Software Development Kit (SDK) module.

The driving behavior data may include driving time information, mileage information, speed information, steering information, latitude and longitude information, altitude information, mobile phone call status information, rapid acceleration information, rapid deceleration information, and sharp turn information, etc. of the target driver.

The server can communicate with a mobile terminal equipped with a mobile SDK module via a network.

Specifically, the mobile SDK module in the mobile terminal of the target driver can be used to collect GPS, accelerometer, and gyroscope sensor data during the driving process of the target driver to obtain the driving behavior data of the target driver and drive the target driver. The driver's driving behavior data is uploaded to the server, and the server binds and stores the driving behavior data of the target driver and the target driver's identity.

The SDK module may be located in a mobile terminal, and the mobile terminal may be a mobile terminal of a target driver, such as a mobile phone. The phone call status information can reflect whether the target driver is on the phone during driving. Because the driver calls while driving is a behavior that hinders safe driving, when considering the driver ’s driving behavior, taking into account the status of the phone call can improve the accuracy of the scoring results.

In the embodiment of the present invention, since the driving behavior data of the target driver is collected based on the SDK module in the mobile terminal of the target driver, it can avoid that multiple drivers can be registered under the same vehicle to a certain extent As a result, the driving records of the same vehicle are not affiliated with the same driver, so the driving behavior scores can be predicted separately for different drivers.

112: Obtain the driving environment data of the target driver through the vehicle depth camera module.

The driving environment related data may include target detection information, ranging information, road condition information, and lane detection information.

The server can communicate with the vehicle depth camera module through the network, and the mobile terminal equipped with the mobile SDK module can communicate with the vehicle depth camera module.

Specifically, the in-vehicle depth camera module may be mounted on a vehicle of a target driver. You can use the in-vehicle depth camera module to perform target detection, ranging, road conditions, and lane detection during driving to collect data about the driving environment of the target driver, and upload the collected data related to the driving environment to the server.

113: Obtain business policy data of the target driver through a preset interface.

The business policy data of the target driver includes the basic information of the target driver, policy purchase information, and claim information. Among them, the policy purchase information includes information such as the type of insurance and the amount of insurance.

Specifically, the preset interface may be an application programming interface (API) or other types of interfaces. The server can interface with the business policy system through a preset interface, and obtain the business policy data corresponding to the target driver's identity from the business policy system according to the identity of the target driver. The side identification can be the target driver's Mobile phone number, user name, or other information that uniquely identifies the driver.

FIG. 3 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention. The embodiment shown in FIG. 3 is an extension of the embodiment shown in FIG. 1. In order to avoid repetition, the differences are highlighted here. Place. As shown in FIG. 3, 120 in FIG. 1 includes the following content.

121: Extract the first characteristic of the target driver from the driving behavior data of the target driver.

The first feature includes at least one of the average driving mileage, the average driving speed, the number of rapid accelerations, the number of rapid decelerations, the number of sharp turns, and whether or not the driver is fatigued.

Specifically, for each target driver, driving time information, mileage information, speed information, steering information, latitude and longitude information, altitude information, mobile phone call status information, rapid acceleration information, and emergency information included in the driving behavior data of the target driver. In the deceleration information and sharp turn information, the monthly or daily average driving mileage, average driving speed, number of rapid accelerations, number of rapid decelerations, number of sharp turns, and fatigue are extracted according to a preset time window and preset mileage. At least one of driving.

Alternatively, a first feature may be extracted from driving behavior data corresponding to each driving trip, that is, a score corresponding to each driving trip.

122: Extract the second characteristic of the target driver from the driving environment data of the target driver.

The second feature includes at least one of a driving road state and a number of lane changes.

Specifically, for each target driver, at least one of a driving road state and a lane changing number is extracted from target detection information, ranging information, road condition information, and lane detection information included in the driving environment data of the target driver.

123: Extract the third characteristic and the fourth characteristic of the target driver from the business policy data of the target driver.

The third feature includes basic personal information, and the fourth feature includes at least one of policy purchase record information and claim history information. Personal basic information may include age, gender, health information, and so on.

124: Combine the first feature, the second feature, the third feature, and the fourth feature to determine target feature data.

Specifically, the first feature, the second feature, the third feature, and the fourth feature may be combined according to the identity of the target driver to obtain target feature data. Since the target characteristic data can more fully represent the driving behavior of the target driver, the accuracy of the scoring results can be improved.

The first feature, the second feature, the third feature, and the fourth feature may be four vectors having the same or different dimensions. The target feature data obtained after the feature combination may also be a vector, and the dimension of the vector may be a feature combination. Sum of the dimensions of the first four vectors.

Optionally, before 120, the method may further include: performing data preprocessing on the driving behavior data, driving environment data, and business insurance data of the target driver, which may specifically include data cleaning and data conversion to facilitate the characteristics of subsequent steps extract.

According to an embodiment of the present invention, the method in FIG. 1 further includes: training the machine learning model using sample data to obtain a driving behavior score model.

FIG. 4 is a schematic flowchart of a driving behavior scoring method according to another embodiment of the present invention. The embodiment shown in FIG. 4 is based on the embodiment shown in FIG. 1. To avoid repetition, the diagram is omitted here. The content described in 1 focuses on the differences. As shown in FIG. 4, the driving behavior scoring method includes the following contents.

140: Obtain sample data. The sample data includes driving behavior data, driving environment data, and business insurance data of the sample driver.

The driving behavior data of the sample driver is obtained through the mobile SDK module, and the driving behavior data of the sample driver is uploaded to the server, and the driving behavior data of the sample driver and the sample driver's identification are bound and stored by the server. Since the driving behavior data of the sample driver is collected based on the SDK module in the sample driver's own mobile terminal, it can avoid to a certain extent that multiple drivers can be registered under the same vehicle and cause the same vehicle. The driving records do not belong to the same driver, which can avoid data interference in the subsequent modeling process of the driving behavior scoring model.

The in-vehicle depth camera module is used to obtain the driving environment data of the sample driver, and upload the driving environment data of the sample driver to the server, and the server binds and stores the relevant data of the driving environment of the sample driver and the identity of the sample driver.

Obtain business policy data of the sample driver through a preset interface.

The specific implementation process of 140, and the information specifically included in the driving behavior data, driving environment data, and business insurance policy data of the sample driver can refer to the description in FIG. 1 and FIG. 2 described above, and will not be repeated here.

150: Determine the sample characteristic data and the sample driver's driving behavior score based on the sample driver's driving behavior data, driving environment data, and business policy data.

Specifically, the sample characteristic data is determined based on the driving behavior data, driving environment data, and business insurance data of the sample driver; the driving behavior score of the sample driver is determined based on the business insurance data of the sample driver.

In an embodiment, determining the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver includes: extracting a fifth characteristic of the sample driver from the driving behavior data of the sample driver; and The sixth characteristic of the sample driver is extracted from the driver's driving environment data; the seventh and eighth characteristics of the sample driver are extracted from the business insurance data of the sample driver; the fifth, sixth, and Seventh feature and eighth feature, determine sample feature data. Determining the driving behavior score of the sample driver based on the business insurance policy data of the sample driver includes determining the driving behavior score of the sample driver based on the seventh feature and the eighth feature.

Specifically, the fifth feature may include at least one of an average driving mileage, an average driving speed, a number of rapid accelerations, a number of rapid decelerations, a number of sharp turns, and whether the driver is fatigued. The sixth feature may include at least one of a driving road state and a number of lane changes. The seventh feature may include personal basic information. The eighth feature may include at least one of policy purchase record information and claim history information.

The fifth feature, the sixth feature, the seventh feature, and the eighth feature are combined according to the identity of the sample driver to obtain sample feature data. The sample feature data of the sample driver may be stored in a list.

For a specific implementation process of 150, reference may be made to the foregoing description in FIG. 1 and FIG. 3, and details are not described herein again.

In the embodiment of the present invention, the sample feature data is obtained by extracting features from the driving behavior data, driving environment data, and business policy data of the sample driver and combining the features, so that the sample feature data obtained can be more fully characterized. Sample the driving behavior of the driver, so that the model obtained from subsequent training can evaluate the driving behavior of the driver more finely.

In an embodiment, determining the driving behavior score of the sample driver based on the seventh feature and the eighth feature includes: calculating a claim rate of the sample driver according to the seventh feature and the eighth feature; and according to a preset mapping relationship table , Determine the driving behavior score of the sample driver that has a mapping relationship with the claim rate, where the higher the claim rate, the lower the driving behavior score of the sample driver.

Specifically, the features extracted from the business insurance policy data include the claim history information of the sample driver, and the claim rate of the sample driver can be calculated based on the claim history information of the sample driver.

In the mapping relationship table, the higher the claim rate is, the lower the driving behavior score is. For example, the score range may be 0 to 100. The driving behavior score obtained by the mapping can be used as the label information of the sample driver.

160: Train the machine learning model according to the sample characteristic data and the driving behavior score of the sample driver to obtain a driving behavior score model.

The trained driving behavior scoring model is stored for scoring the driving behavior of the target driver other than the sample driver. 140 to 160 can be performed before 130.

According to an embodiment of the present invention, the machine learning model is a tree-based integrated learning model XGBoost (eXtreme Gradient Boosting). For a given data set D = {(x _i , y _i )}, the XGBoost model function form is as follows:

Where K is the number of trees to be learned, and x _i is the input,

Represents the prediction result. F is the hypothesis space, and f (x) is the classification and regression tree CART (Classification and Regression Tree):

F = {f (x) = w _{q (x)} } (q: R ^m → T, w∈R ^T )

Among them, q (x) indicates that the sample x is assigned to a certain leaf node, and w is the score of the leaf node, so w _{q (x)} represents the predicted value of the regression tree on the sample. From the above XGBoost model function, it can be seen that the model uses the prediction results of each regression tree in the K regression trees to iteratively calculate to obtain the final prediction result.

Moreover, the input samples of each regression tree are related to the training and prediction of the previous regression tree.

Specifically, the feature data of the sample driver is used as the x _i in the data set D, and the driving behavior score corresponding to the feature data of the sample driver is used as the y _i in the data set D to learn the K regression in the XGBoost model. Tree parameters, that is, determining the input x _i and its output of each regression tree

, Where x _i can be an n-dimensional vector or array. That is, by inputting known training sample data x _i , the prediction result of the above model is

Compare with the actual mapping label y _{i of the} training sample data, and continuously adjust the model parameters until the expected accuracy is reached, determine the model parameters, thereby establishing a driving behavior scoring model, and store the driving behavior scoring model offline for online use Called when driving behavior is scored.

In addition, the machine learning model may also be a boosting tree model other than the XGBoost model, or may be another type of machine learning model, such as a random forest model, which is not limited in the present invention.

The driving behavior scoring method provided by the embodiment of the present invention starts from the driving behavior data, driving environment data, and business policy data of a sample driver, extracts sample characteristic data, and obtains samples based on the sample driver's business policy data. The driver's driving behavior score, and training the machine learning model based on the sample characteristic data and the sample driver's driving behavior score to obtain a driving behavior scoring model, thereby achieving multi-dimensional and refined modeling of the driving behavior of the driver In addition, the driving behavior scoring model can be used to accurately predict the driving behavior scoring for different drivers.

FIG. 5 is a schematic structural diagram of a driving behavior scoring device 500 according to an embodiment of the present invention. As shown in FIG. 5, the apparatus 500 includes an obtaining module 510, a determining module 520, and a scoring module 530.

The obtaining module 510 is used to obtain driving behavior data, driving environment data, and business insurance policy data of the target driver; the determination module 520 is used to determine target feature data based on the driving behavior data, driving environment data, and business insurance data of the target driver; a scoring module 530 is used to input target feature data into a driving behavior scoring model to obtain a driving behavior score of the target driver.

An embodiment of the present invention provides a driving behavior scoring device, which extracts target characteristic data of a target driver from multiple dimensions of the target driver's driving behavior data, driving environment data, and business policy data, and uses the driving behavior score The model predicts the driving behavior score of the target driver, which can improve the accuracy of the scoring results, and then improve the accuracy of auto insurance pricing.

According to an embodiment of the present invention, the obtaining module 510 is configured to obtain driving behavior data of a target driver through a mobile software development kit module; obtain driving environment data of the target driver through a vehicle depth camera module; and obtain target driving through a preset interface. Employee's business policy data.

According to an embodiment of the present invention, the determining module 520 is configured to extract a first characteristic of the target driver from the driving behavior data of the target driver, and the first characteristic includes an average driving mileage, an average driving speed, a number of rapid accelerations, and a rapid deceleration. At least one of the number of times, the number of sharp turns, and the fatigue driving; the second characteristic of the target driver is extracted from the driving environment data of the target driver, and the second characteristic includes at least one of the driving state and the number of lane changes; The third and fourth characteristics of the target driver are extracted from the target driver's business policy data. The third characteristic includes basic personal information, and the fourth characteristic includes at least one of policy purchase record information and claim history information. Feature, second feature, third feature, and fourth feature to determine target feature data.

According to an embodiment of the present invention, the apparatus 500 further includes: a training module 540, configured to train a machine learning model using sample data to obtain a driving behavior scoring model.

According to an embodiment of the present invention, the training module 540 is configured to: obtain sample data, the sample data including driving behavior data, driving environment data, and business insurance policy data of the sample driver; based on the driving behavior data, driving environment data, and business of the sample driver The policy data determines the sample characteristic data and the driving behavior score of the sample driver; the machine learning model is trained based on the sample characteristic data and the driving behavior score of the sample driver to obtain the driving behavior scoring model.

According to an embodiment of the present invention, the training module 540 is configured to determine the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver; determine the driving behavior score of the sample driver based on the business insurance data of the sample driver .

According to an embodiment of the present invention, the training module 540 is configured to: extract the fifth feature of the sample driver from the driving behavior data of the sample driver; extract the sixth feature of the sample driver from the driving environment data of the sample driver Extract the seventh and eighth characteristics of the sample driver from the business insurance data of the sample driver; combine the fifth, sixth, seventh, and eighth characteristics to determine the sample characteristic data; based on the seventh characteristic And the eighth feature determines the driving behavior score of the sample driver.

According to an embodiment of the present invention, the training module 540 is configured to calculate a claim rate of the sample driver according to the seventh feature and the eighth feature; and determine a sample driver having a mapping relationship with the claim rate according to a preset mapping relationship table. The driving behavior score of the sample driver is higher, the higher the claim rate is, the lower the driving behavior score of the sample driver is.

According to an embodiment of the present invention, the machine learning model is a lifting tree model or a random forest model.

The driving behavior scoring device provided by the embodiment of the present invention belongs to the same inventive concept as the driving behavior scoring method provided by the embodiment of the present invention, and can execute the driving behavior scoring method provided by any embodiment of the present invention, and is equipped with a corresponding method for performing driving behavior scoring Functional modules and beneficial effects. For technical details not described in detail in this embodiment, reference may be made to the driving behavior scoring method provided in the embodiment of the present invention, and details are not described herein again.

FIG. 6 is a block diagram of a computer device 60 according to an embodiment of the present invention, which includes: one or more processors 61; a memory 62, configured to store one or more programs; Each processor 61 executes such that one or more processors 61 implement the driving behavior scoring method described in the above embodiment.

In addition, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the driving behavior scoring method according to the foregoing embodiment is implemented.

Those skilled in the art should understand that the embodiments in the embodiments of the present invention may be provided as a method, a system, or a computer program product. Therefore, the embodiments of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. .

The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, and combinations of processes and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, so that the instructions generated by the processor of the computer or other programmable data processing device are used to generate instructions Means for implementing the functions specified in one or more flowcharts and / or one or more blocks of the block diagrams.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to work in a particular manner such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction device, the instructions The device implements the functions specified in one or more flowcharts and / or one or more blocks of the block diagram.

These computer program instructions can also be loaded on a computer or other programmable data processing device, so that a series of steps can be performed on the computer or other programmable device to produce a computer-implemented process, which can be executed on the computer or other programmable device. The instructions provide steps for implementing the functions specified in one or more flowcharts and / or one or more blocks of the block diagrams.

Although the preferred embodiments of the embodiments of the present invention have been described, those skilled in the art can make other changes and modifications to these embodiments once they know the basic inventive concepts. Therefore, the following claims are intended to be construed to include the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the invention.

Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. In this way, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention also intends to include these modifications and variations.

Claims

A driving behavior scoring method, comprising:

Obtain the driving behavior data, driving environment data, and business policy data of the target driver;

Determining target characteristic data based on the driving behavior data, driving environment data, and business insurance data of the target driver;

The target characteristic data is input into a driving behavior scoring model to obtain a driving behavior score of the target driver.
The method according to claim 1, wherein the acquiring the driving behavior data, the driving environment data, and the business insurance policy data of the target driver comprises:

Obtaining the driving behavior data of the target driver through a mobile software development kit module;

Acquiring the driving environment data of the target driver through a vehicle-mounted depth camera module;

The business insurance data of the target driver is acquired through a preset interface.
The method according to claim 1 or 2, wherein determining the target characteristic data based on the driving behavior data, driving environment data, and business insurance data of the target driver comprises:

A first characteristic of the target driver is extracted from the driving behavior data of the target driver, and the first characteristic includes average driving mileage, average driving speed, number of rapid accelerations, number of rapid decelerations, number of sharp turns, and whether At least one of fatigue driving;

Extracting a second characteristic of the target driver from the driving environment data of the target driver, the second characteristic including at least one of a driving road state and a number of lane changes;

A third characteristic and a fourth characteristic of the target driver are extracted from the target driver's business policy data, the third characteristic includes basic personal information, and the fourth characteristic includes policy purchase record information and claim history At least one of the information;

The first feature, the second feature, the third feature, and the fourth feature are combined to determine the target feature data.
The method according to any one of claims 1 to 3, further comprising:

A machine learning model is trained using sample data to obtain the driving behavior scoring model.
The method according to claim 4, wherein the training a machine learning model using sample data to obtain the driving behavior scoring model comprises:

Acquiring the sample data, the sample data including driving behavior data, driving environment data, and business insurance data of a sample driver;

Determining sample feature data and driving behavior scores of the sample driver based on the driving behavior data, driving environment data, and business insurance data of the sample driver;

Training the machine learning model according to the sample characteristic data and the driving behavior score of the sample driver to obtain the driving behavior score model.
The method according to claim 5, wherein determining the sample characteristic data and the driving behavior score of the sample driver based on the driving behavior data, driving environment data, and business insurance data of the sample driver comprises:

Determining the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver;

A driving behavior score of the sample driver is determined based on the business insurance data of the sample driver.
The method according to claim 6, wherein the determining the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver comprises:

Extracting a fifth feature of the sample driver from the driving behavior data of the sample driver;

Extracting a sixth feature of the sample driver from the driving environment data of the sample driver;

Extracting the seventh characteristic and the eighth characteristic of the sample driver from the business insurance data of the sample driver;

Combining the fifth feature, the sixth feature, the seventh feature, and the eighth feature to determine the sample feature data; wherein,

The determining the driving behavior score of the sample driver based on the business insurance policy data of the sample driver includes:

A driving behavior score of the sample driver is determined based on the seventh feature and the eighth feature.
The method according to claim 7, wherein determining the driving behavior score of the sample driver based on the seventh feature and the eighth feature comprises:

Calculating the claim rate of the sample driver according to the seventh feature and the eighth feature;

A driving behavior score of the sample driver having a mapping relationship with the claim rate is determined according to a preset mapping relationship table, wherein the higher the claim rate is, the lower the driving behavior score of the sample driver is.
The method according to any one of claims 4 to 8, wherein the machine learning model is a boosted tree model or a random forest model.
A driving behavior scoring device, comprising:

An acquisition module for acquiring driving behavior data, driving environment data, and business policy data of a target driver;

A determining module, configured to determine target characteristic data based on the driving behavior data, driving environment data, and business insurance data of the target driver;

A scoring module is configured to input the target characteristic data into a driving behavior scoring model to obtain a driving behavior score of the target driver.
The apparatus according to claim 10, wherein the obtaining module is configured to:

Obtaining the driving behavior data of the target driver through a mobile software development kit module;

Acquiring the driving environment data of the target driver through a vehicle-mounted depth camera module;

The business insurance data of the target driver is acquired through a preset interface.
The apparatus according to claim 10 or 11, wherein the determining module is configured to:

A first characteristic of the target driver is extracted from the driving behavior data of the target driver, and the first characteristic includes average driving mileage, average driving speed, number of rapid accelerations, number of rapid decelerations, number of sharp turns, and whether At least one of fatigue driving;

Extracting a second characteristic of the target driver from the driving environment data of the target driver, the second characteristic including at least one of a driving road state and a number of lane changes;

A third characteristic and a fourth characteristic of the target driver are extracted from the target driver's business policy data, the third characteristic includes basic personal information, and the fourth characteristic includes policy purchase record information and claim history At least one of the information;

The first feature, the second feature, the third feature, and the fourth feature are combined to determine the target feature data.
The device according to any one of claims 10 to 12, further comprising:

A training module is used to train a machine learning model using sample data to obtain the driving behavior scoring model.
The apparatus according to claim 13, wherein the training module is configured to:

Acquiring the sample data, the sample data including driving behavior data, driving environment data, and business insurance data of a sample driver;

Determining sample feature data and driving behavior scores of the sample driver based on the driving behavior data, driving environment data, and business insurance data of the sample driver;

Training the machine learning model according to the sample characteristic data and the driving behavior score of the sample driver to obtain the driving behavior score model.
The apparatus according to claim 14, wherein the training module is configured to:

Determining the sample characteristic data based on the driving behavior data, driving environment data, and business insurance data of the sample driver;

A driving behavior score of the sample driver is determined based on the business insurance data of the sample driver.
The apparatus according to claim 15, wherein the training module is configured to:

Extracting a fifth feature of the sample driver from the driving behavior data of the sample driver;

Extracting a sixth feature of the sample driver from the driving environment data of the sample driver;

Extracting the seventh characteristic and the eighth characteristic of the sample driver from the business insurance data of the sample driver;

Combining the fifth feature, the sixth feature, the seventh feature, and the eighth feature to determine the sample feature data;

A driving behavior score of the sample driver is determined based on the seventh feature and the eighth feature.
The apparatus according to claim 16, wherein the training module is configured to:

Calculating the claim rate of the sample driver according to the seventh feature and the eighth feature;

A driving behavior score of the sample driver having a mapping relationship with the claim rate is determined according to a preset mapping relationship table, wherein the higher the claim rate is, the lower the driving behavior score of the sample driver is.
The device according to any one of claims 13 to 17, wherein the machine learning model is a lifting tree model or a random forest model.
A computer device, comprising:

One or more processors;

Memory for storing one or more programs;

When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the driving behavior scoring method according to any one of claims 1 to 9.
A computer-readable storage medium having stored thereon a computer program, characterized in that when the program is executed by a processor, the driving behavior scoring method according to any one of claims 1 to 9 is implemented.