Order value stratification method, computer-readable storage medium, and computer device

Abstract

The present invention is applicable to the logistics field and provides an order value stratification method, a computer-readable storage medium, and a computer device, including obtaining a historical transaction data sample of an order; modeling based on the historical transaction data sample of the order and selecting relevant features of each order; inputting the relevant features into a machine learning model, and the machine learning model sorts the importance of the relevant features according to the size of the gain value with the completion of the order as the goal; screening important features as the input of a regression model according to the importance ranking of the relevant features, and outputting the weight values corresponding to the important features with the completion of the order as the goal; calculating the feature values of the important features, multiplying the feature values by the weight values corresponding to the important features, and summing the products to obtain an order score; sorting according to the order score and dividing the orders into multiple layers according to the sorting. This makes the order value not affected by the current supply and demand and the driver level.

Description

Technical Field

[0001] This application belongs to the field of logistics, and particularly relates to an order value stratification method, a computer-readable storage medium, and a computer device. Background Art

[0002] Currently, the basic order value evaluation method is only a post-method. That is, after an order is generated, as time goes by, it is gradually broadcast and pushed to drivers within different ranges. Generally, the farther a driver is from the order, the later they receive the order. The shorter the driver's order acceptance time, the faster the order is responded to, indicating that the order has a higher value in the market. Or the more people participate in the PK for an order, the higher the order value.

[0003] Due to the above situation being severely affected by real-time supply and demand, in the case where the driver fails to fulfill the contract, it is currently impossible to calculate the probability that the order can be fulfilled. Moreover, the levels of drivers on the platform vary, with both experienced drivers with high fulfillment rates and new drivers with low fulfillment rates. The order value is affected by the current supply and demand and the driver level. Summary of the Invention

[0004] The purpose of the present invention is to provide an order value stratification method, a computer-readable storage medium, and a computer device, aiming to solve the problem that the order value is affected by the current supply and demand and the driver level.

[0005] In a first aspect, the present invention provides an order value stratification method, including:

[0006] Obtain historical transaction data samples of the order;

[0007] Build a model based on the historical transaction data samples of the order, and select relevant features of each order;

[0008] Input the relevant features into a machine learning model. The machine learning model aims at the completion of the order based on the relevant features, and sorts the relevant features according to the size of the gain value;

[0009] Screen important features as the input of the regression model according to the importance ranking of the relevant features, and output the weight values corresponding to the important features with the aim of the completion of the order; the gain value of the important features is greater than a predetermined gain value;

[0010] Calculate the feature values of the important features, multiply the feature values by the weight values corresponding to the important features, and sum the products to obtain the order score;

[0011] Sort according to the order scores, and divide the orders into multiple layers according to the sorting.

[0012] Further, the method further includes:

[0013] Calculate the fulfillment rate of all orders in each layer and evaluate the gap in the fulfillment rate between layers; the fulfillment rate = the number of completed orders / the number of responded orders;

[0014] Monitor the gap in the fulfillment rate between layers. If it does not match the expectation, adjust the multiple regression model; the expectation is that the fulfillment rate of good orders is higher than that of bad orders.

[0015] Further, the order value stratification method further includes adjusting the stratification of orders according to the city, vehicle type, and actual application scenario.

[0016] Further, the machine learning model includes the XG - Boost model and the classification model.

[0017] Further, the regression model includes the multiple regression model and the logistic regression model.

[0018] Further, the historical transaction data samples of the orders include: order start point features, order end point features, order note features, order price - mileage features, and dispatching shipper user features.

[0019] Further, the relevant features include: order geographical attributes, shipper attributes, note attributes, and price - mileage.

[0020] Further, the order geographical attributes include: order start point features and order end point features.

[0021] Further, the order start point features include the start fulfillment rate, start response rate, and start pairing rate; the order end point features include the end fulfillment rate, end response rate, and end pairing rate.

[0022] In a second aspect, the present invention provides a computer - readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the order value stratification method described in the first aspect are implemented.

[0023] In a third aspect, the present invention provides a computer device, including: one or more processors, a memory, and one or more computer programs, where the processor and the memory are connected through a bus, and the one or more computer programs are stored in the memory and configured to be executed by the one or more processors, and when the processor executes the computer program, the steps of the order value stratification method described in the first aspect are implemented.

[0024] In the present invention, order value stratification is performed using order scores, enabling the determination of the value of an order in the trading market before the order is issued during the order-driver matching process. Based on the order quality, relative inclination of order resources and driver compensation strategies can be implemented during the order matching process, improving the operational space in the order-driver transaction process and enhancing the fairness of the platform's order allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flowchart of an order value stratification method provided by an embodiment of the present invention.

[0026] Figure 2 is a table for evaluating the order fulfillment difficulty based on historical order remarks provided by an embodiment of the present invention.

[0027] Figure 3 is a table of order fulfillment performance after equal stratification based on the order value model evaluation provided by an embodiment of the present invention.

[0028] Figure 4 is a specific structural block diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] In order to make the objectives, technical solutions and beneficial effects of the present invention clearer and more understandable, the following further details the present application in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0030] To illustrate the technical solutions described in the present application, the following is illustrated through specific embodiments.

[0031] Please refer to Figure 1 , an order value stratification method provided by an embodiment of the present invention includes the following steps: It should be noted that if there are substantially the same results, the order value stratification method of the present application is not limited to Figure 1 the process sequence shown.

[0032] S1. Obtain a historical transaction data sample of the order;

[0033] S2. Build a model based on the historical transaction data sample of the order and select relevant features for each order;

[0034] S3. Input the relevant features into a machine learning model. The machine learning model aims at the completion of the order based on the relevant features and sorts the importance of the relevant features according to the magnitude of the gain value;

[0035] S4. Sort and screen important features according to the importance ranking of relevant features as the input of the regression model, and take the completion of the order as the goal to output the weight values corresponding to the important features; the gain value of the important features is greater than the predetermined gain value;

[0036] S5. Calculate the feature values of the important features, multiply the feature values by the weight values corresponding to the important features, and sum the products to obtain the order score;

[0037] S6. Sort according to the order scores and divide the orders into multiple layers according to the sorting.

[0038] In an embodiment of the present invention, the method further includes:

[0039] Calculate the fulfillment rate of all orders in each layer and evaluate the gap in fulfillment rate between layers; the fulfillment rate = the number of completed orders / the number of responded orders;

[0040] Monitor the gap in fulfillment rate between layers. If it does not match the expectation, adjust the multiple regression model; the expectation is that the fulfillment rate of good orders is higher than that of bad orders.

[0041] In an embodiment of the present invention, the machine learning model includes an XG - Boost model and a classification model.

[0042] In an embodiment of the present invention, the regression model includes a multiple regression model and a logistic regression model.

[0043] In an embodiment of the present invention, the order value stratification method further includes adjusting the stratification of orders according to the city, vehicle type, and actual application scenario.

[0044] For example: The order scores can be evenly divided into X layers from high to low, and the highest - scoring layer is defined as layer 1, and the lowest - scoring layer is defined as layer X. Observe the fulfillment rate of each layer and re - divide it into Y layers for easy business understanding.

[0045] Figure 3 It can be verified that the gap in fulfillment rate between layers meets the assumption. Thus, according to the gap between adjacent layers, a coarser - grained stratification can be determined, which can be divided into 3∶5∶2 according to data observation to evaluate the goodness, medium, and badness of orders (that is, the difficulty, medium, and ease of fulfillment).

[0046] When adjusting the score intervals of the stratification according to different cities, vehicle types, and actual application scenarios, the regression model will calculate different order scores and divide the stratification according to a fixed score interval; or as time goes by and the order business gradually changes, when the proportion of orders in a certain layer becomes low, the order stratification model needs to be updated.

[0047] In an embodiment of the present invention, the historical transaction data sample of the order includes: order start point features, order end point features, order note features, order price mileage features, and dispatching shipper user features.

[0048] In an embodiment of the present invention, the relevant features include: order geographical attributes, shipper attributes, note attributes, and price mileage.

[0049] In an embodiment of the present invention, the order geographical attributes include: order start point features and order end point features.

[0050] In an embodiment of the present invention, the order start point features include start point fulfillment rate, start point response rate, and start point matching rate; the order end point features include end point fulfillment rate, end point response rate, and end point matching rate.

[0051] In the order scenario, the note is also a business scenario reflecting the order difficulty. The relevant data of the note character splitting such as Figure 2 , Figure 2 shows a table for evaluating the order implementation difficulty based on the historical order note characters. Then, the fulfillment rate of the relevant orders where a certain character has appeared in the note can be calculated. For orders with notes, the average value of the fulfillment rates of all note characters is taken as the note prediction fulfillment rate.

[0052] According to the calculated eigenvalue of the important feature, multiply the eigenvalue by the weight value corresponding to the important feature, and sum the products to obtain the order score. For example: The finally determined 5 important features are: the fulfillment rate of the start point in the past 28 days, the fulfillment rate of the end point in the past 28 days, the fulfillment rate of the user in the past 28 days, the predicted fulfillment rate of the note, and the order mileage. Then:

[0053] Order score = 0.69 * fulfillment rate of the start point in the past 28 days + 0.18 * fulfillment rate of the end point in the past 28 days + 0.24 * predicted fulfillment rate of the note + 0.22 * fulfillment rate of the user in the past 28 days - 1.4 * 10 ﹣5 * order mileage

[0054] This order value model calculation method can be replicated to each city and vehicle type, and it varies slightly according to the order score distribution of each city. The layering method can be adjusted according to the actual application scenario to achieve the practicality of the city and vehicle type.

[0055] An embodiment of the present invention provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the order value layering method provided in an embodiment of the present invention.

[0056] Figure 4The specific structural block diagram of a computer device provided by an embodiment of the present application is shown. A computer device 100 includes: one or more processors 101, a memory 102, and one or more computer programs. The processor 101 and the memory 102 are connected through a bus. The one or more computer programs are stored in the memory 102 and are configured to be executed by the one or more processors 101. When the processor 101 executes the computer program, the steps of the order value stratification method provided by an embodiment of the present application are implemented.

[0057] The computer device includes a server, a terminal, etc. The computer device may be a desktop computer, a mobile terminal or a vehicle-mounted device. The mobile terminal includes at least one of a mobile phone, a tablet computer, a personal digital assistant or a wearable device, etc.

[0058] In the embodiment of the present invention, the orders are stratified by value using the order score, so that during the matching process of the order and the driver, the value of the order in the trading market can be determined before the order is issued; according to the quality of the order, thus during the order matching process, relative inclination of order resources and driver compensation strategies can be carried out, improving the operation space of the order and the driver during the transaction process, and enhancing the fairness of the platform for order allocation.

[0059] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. The storage medium may include: a read-only memory (ROM, Read Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disc, etc.

[0060] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. An order value stratification method, characterized in that, it includes: Obtain a historical transaction data sample of the order; Build a model based on the historical transaction data sample of the order, and select relevant features of each order; The historical transaction data sample of the order includes: order start point features, order end point features, order note features, order price mileage features, and dispatching shipper user features; the relevant features include: order geographical attributes, shipper attributes, note attributes, and price mileage; the order geographical attributes include: order start point features and order end point features, and the order start point features include start point fulfillment rate, start point response rate, and start point pairing rate; the order end point features include end point fulfillment rate, end point response rate, and end point pairing rate; Input the relevant features into a machine learning model. The machine learning model takes the completion of the order as the goal according to the relevant features, and sorts the importance of the relevant features according to the size of the gain value, and outputs the important features selected by sorting the importance of the relevant features; Take the selected important features as the input of the regression model, and take the completion of the order as the goal, and output the weight values corresponding to the important features; the gain value of the important features is greater than the predetermined gain value; Calculate the feature values of the important features, multiply the feature values by the weight values corresponding to the important features, and sum the products to obtain the order score; the feature value of the note attribute includes the note predicted fulfillment rate; Sort according to the order score, and divide the orders into multiple layers according to the sorting.

2. The order value stratification method according to claim 1, characterized in that, the method further includes: Calculate the fulfillment rate of all orders in each layer, and evaluate the gap in fulfillment rate between layers; the fulfillment rate = the number of completed orders / the number of responded orders; Monitor the gap in fulfillment rate between layers. If it does not match the expectation, adjust the multiple regression model; the expectation is that the fulfillment rate of good orders is higher than that of bad orders.

3. The order value stratification method according to claim 1, characterized in that, the machine learning model includes an XG-Boost model and a classification model.

4. The order value stratification method according to claim 1, characterized in that, the regression model includes a multiple regression model and a logistic regression model.

5. The order value stratification method according to claim 1, characterized in that, the order value stratification method further includes adjusting the stratification of the order according to the city, vehicle type, and actual application scenario.

6. A computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, it implements the steps of the order value stratification method according to any one of claims 1 to 5.

7. A computer device, including: One or more processors, a memory, and one or more computer programs. The processor and the memory are connected by a bus. Among them, the one or more computer programs are stored in the memory and are configured to be executed by the one or more processors. Characterized in that when the processor executes the computer program, it implements the steps of the order value stratification method according to any one of claims 1 to 5.

Images