Research On Bidding Strategy Of NTOCC Platform

With the development of the Internet and logistics industry,the relationship between them is becoming closer.Due to the high requirements on timeliness and reliability in terms of logistics information circulation and resource scheduling integration,the innovation model of logistics development combined with the Internet has also changed from multiple directions One after another emerged,and the carless carrier platform was born.This logistics service model can greatly reduce the distance between goods and people,save the transaction communication costs between shippers and carriers,and improve the efficiency of urban logistics services.This article is based on an in-depth analysis of the platform business of the carless carrier.It falls on the quotation behavior of the carrier on the bidding order,combines the product auction theory and the bidding game theory similar to the logistics bidding behavior,and innovatively uses the game model and Data mining is analyzed in two directions.Some research results have been obtained in explaining the influencing factors of the carrier's quotation behavior,the user group of the carrier and the quotation prediction.In terms of game models,based on auction and game theory,this paper uses a static Bayesian game model to model the optimal bidding strategy of the merchant,and initially obtains the single freight cost and carrier bidding behavior from the perspective of the game model.Preference and estimation of competitors' bid distribution are the conclusions of three key factors that affect the carrier's optimal bid in the process of participating in platform bidding.In addition,this paper starts with the data of carrier quotes from the carless carrier platform,and we apply data mining algorithms for interpretive and predictive effects to complete three parts of research.First of all,the first part is to use K-means clustering algorithm to classify users based on user value,stickiness,and carrying capacity,and compare the differences between carriers based on different groups.At the same time,the clustering result label corresponding to each carrier in this part can play a certain carrier feature effect in the later research.Then,the second part is to explore the factors that may affect the carrier's quotation,and also to verify the conclusion of the game model part from the perspective of data.The results show that freight costs,bidding behavior preferences,the number of people participating in the competition and the distribution of quotes have a significant impact on the quotation.Finally,based on the carrier's quotation data,this paper trains a model that can predict the carrier's appropriate quotation,and tries to optimize the prediction effect of the model from linear model to random forest and nonlinear model of LightGBM to Stacking fusion model.The prediction accuracy rate is only 7%different from the average quote price.This section aims to provide a suitable reference price for car owners when quoting.