Construction of Driver Credit Evaluation System Based on Driver Behavior Data from Online Freight Platform
The market size of online freight platforms in China is growing rapidly,and the level of digitali-zation,standardization,and intensification of the road freight transportation industry is constantly rising.However,during transportation,due to the bad habits of some drivers which cause poor service quality and high transportation risks,the platforms and cargo owners have low credit ratings of the drivers,giving promi-nence to the contradiction between the rapid development of the online freight platforms and the imperfect driver credit rating system.Constraining the credit level of platform drivers based on their driving behavior can reduce transportation risks,ensure the safety of the drivers,the cargoes,and the overall road traffic pro-cess,and facilitate the adoption of targeted measures for truck drivers with different credit levels to encourage them to improve service quality and transportation efficiency.First,the driver behavioral profile is calculated based on the driver's waybill trajectory,driving speed,fatigue level and stop data.Common factors including transportation style,transportation risk,transportation efficiency and transportation preference are extracted through factor analysis to establish a driver credit evaluation system.Secondly,the driver's driving credit lev-el is divided based on the SOM-k-means algorithm,and 14 driver types are obtained.From the perspective of service quality and risk level,the 14 types are further summarized into 9 credit types such as the efficient and stable type,mature and reliable type,and efficient and exploratory type,etc.Thirdly,a driver credit scor-ing model is constructed using the random forest algorithm,which converts the characteristic values of actual carriers into corresponding scores,ranking the key factors affecting the driver's credit evaluation in terms of importance,and grouping the driver's credit level into three categories of high value,medium value and low value by calculating the characteristic score.Finally,governance suggestions based on the driver credit of the online freight platforms are put forward,including:attracting"high value"drivers to join the platform,en-couraging"medium value drivers"to transform to high value,and removing drivers falling into the"low val-ue driver"group consecutively.Assessing the credit rating based on drivers'driving behavior can urge the drivers to drop bad driving habits,reduce transportation risks,improve transportation efficiency,further im-prove the service level of online freight platforms,and provide decision-making basis for the platforms and relevant regulatory authorities.
NETL
NSTL
万方数据
