Logistics on-time guarantee service recommendation method, device, computer equipment and storage medium

Abstract

The present application relates to a method, apparatus, computer equipment, and storage medium for recommending the above-mentioned logistics on-time guarantee service, wherein the method includes: obtaining real-time order data of the current order, extracting logistics features from the real-time order data, and inputting the extracted logistics features into a preset on-time delivery discrimination model; if the current order is an on-time delivery type order, obtaining the customer features of the current order customer; and recommending the on-time guarantee service when the current order customer is a target customer for the on-time guarantee service. Throughout the entire process, based on two discriminations based on the preset on-time delivery discrimination model and the current order customer features, the target customers for the on-time guarantee service can be accurately found, avoiding the inconvenience caused to customers by large-scale, aimless promotion, and significantly improving the efficiency of recommending the on-time guarantee service.

Description

Technical Field

[0001] The present application relates to the field of big data technology, and in particular to a method, device, computer equipment, and storage medium for recommending on-time logistics services. Background Art

[0002] With the rapid development of the logistics industry, competition among peers has become increasingly fierce, and differentiated and refined services have become increasingly important. In this context, many companies have launched personalized on-time guarantee services to meet the service needs of different users.

[0003] Currently, there are two main methods for implementing on-time service among industries: 1. Some companies adopt a crude method of pushing default services to all customers or forced binding, ignoring the user experience of customers with low timeliness requirements; 2. Some companies will push on-time guarantee products to designated business areas based on the characteristics of their own business rules, resulting in low service coverage.

[0004] For businesses, on-time delivery guarantee services are a valuable tool for both enriching their business and increasing profits, making their implementation undeniable. However, existing implementation methods require couriers to widely recommend this service and guide customers through the terminal. This can easily lead to user frustration and lack of targetedness, potentially leading to customer churn or revenue loss. Furthermore, recommendations are inefficient, causing inconvenience for both couriers and customers. Overall, the industry lacks an effective recommendation program for on-time delivery guarantee services. Summary of the Invention

[0005] Based on this, it is necessary to provide an effective and efficient logistics on-time guarantee service recommendation method, device, computer equipment and storage medium to address the above technical problems.

[0006] A method for recommending logistics on-time guarantee services, the method comprising:

[0007] Get real-time order data for the current order;

[0008] Extract logistics features from real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model;

[0009] If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer;

[0010] When the current order customer is identified as a target customer for the on-time guarantee service based on customer characteristics, the on-time guarantee service is recommended.

[0011] In one embodiment, before extracting logistics features from real-time order data and inputting the extracted logistics features into a preset on-time delivery discrimination model, the process further includes:

[0012] Get order data of historical orders;

[0013] Extract metadata features that affect delivery timeliness from historical order data;

[0014] The extracted metadata features are input into the initial logistic regression model for training to obtain a preset on-time delivery discrimination model. The category division threshold of the initial logistic regression model is the preset probability threshold, which is related to the historical order timeliness achievement rate.

[0015] In one embodiment, the extracted metadata features are input into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model including:

[0016] Inputting the extracted metadata features into a plurality of different initial logistic regression models for training to obtain a plurality of trained initial logistic regression models, wherein the number of iterations and regularization coefficients in the different initial logistic regression models are different;

[0017] Acquire preset sample test data, and test the multiple trained initial logistic regression models according to the preset sample test data to obtain test results of the multiple trained initial logistic regression models;

[0018] Based on the test results and the preset model selection indicators, the optimal model is selected from multiple trained initial logistic regression models as the preset on-time delivery discrimination model.

[0019] In one embodiment, inputting the extracted metadata features into an initial logistic regression model for training includes:

[0020] The chi-square test method is used to verify the correlation between the extracted metadata features and whether the order is delivered on time;

[0021] Metadata features with greater correlation are selected and input into the initial logistic regression model for training.

[0022] In one embodiment, selecting metadata features with high correlation and inputting them into an initial logistic regression model for training includes:

[0023] The metadata features with large correlation are normalized by using the average coding method to obtain the normalized features;

[0024] The normalized features are input into the initial logistic regression model for training.

[0025] In one embodiment, the above-mentioned logistics on-time guarantee service recommendation method further includes:

[0026] Obtain seed customers who have purchased the on-time insurance service in the historical records;

[0027] Extract customer characteristics of seed customers;

[0028] Based on the customer characteristics of the current order customer and the customer characteristics of the seed customer, determine whether the current order customer is the target customer of the on-time insurance business.

[0029] In one embodiment, judging whether the current order customer is a target customer for the on-time guarantee service based on the customer characteristics of the current order customer and the customer characteristics of the seed customer includes:

[0030] Calculate the cosine similarity between the current order customer and the seed customer;

[0031] Based on the cosine similarity, determine whether the current order customer is the target customer of the on-time guarantee service.

[0032] A logistics on-time guarantee service recommendation device, comprising:

[0033] Order acquisition module, used to obtain real-time order data of the current order;

[0034] Logistics feature extraction module, used to extract logistics features from real-time order data and input the extracted logistics features into the preset on-time delivery discrimination model;

[0035] The customer feature extraction module is used to obtain the customer features of the current order customer when the current order belongs to the on-time delivery type order;

[0036] The service recommendation module is used to recommend the on-time guarantee service when it is identified that the current order customer is a target customer of the on-time guarantee service based on customer characteristics.

[0037] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are performed:

[0038] Get real-time order data for the current order;

[0039] Extract logistics features from real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model;

[0040] If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer;

[0041] When the current order customer is identified as a target customer for the on-time guarantee service based on customer characteristics, the on-time guarantee service is recommended.

[0042] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0043] Get real-time order data for the current order;

[0044] Extract logistics features from real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model;

[0045] If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer;

[0046] When the current order customer is identified as a target customer for the on-time guarantee service based on customer characteristics, the on-time guarantee service is recommended.

[0047] The aforementioned method, apparatus, computer device, and storage medium for recommending on-time logistics services obtains real-time order data for the current order, extracts logistics features from the real-time order data, and inputs these features into a pre-set on-time delivery discrimination model. If the current order is an on-time delivery order, the customer features of the current order are obtained. If the current order customer is a target customer for the on-time delivery service, the on-time delivery service is recommended. This dual process, based on the pre-set on-time delivery discrimination model and the current order customer features, accurately identifies target customers for the on-time delivery service, avoids the inconvenience caused to customers by widespread, untargeted promotions, and significantly improves the efficiency of on-time delivery service recommendations. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a diagram of an application environment of a method for recommending on-time logistics services in one embodiment;

[0049] Figure 2 A flowchart of a method for recommending on-time logistics services in one embodiment is shown;

[0050] Figure 3 A flowchart of a method for recommending on-time logistics services in another embodiment;

[0051] Figure 4 A flowchart of a method for recommending on-time logistics services in another embodiment is shown;

[0052] Figure 5 A schematic diagram of the process for mining target customers who can be recommended for on-time insurance;

[0053] Figure 6 This is a diagram of the technical principle architecture of the recommended method for applying for logistics on-time guarantee business;

[0054] Figure 7 This is a structural block diagram of a device for recommending on-time logistics services in one embodiment;

[0055] Figure 8 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0057] The logistics on-time guarantee service recommendation method provided by this application can be applied to Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. Terminal 102 sends current order data to server 104; server 104 is pre-loaded with a preset on-time delivery discrimination model. Server 104 obtains real-time order data for the current order, extracts logistics features from the real-time order data, and inputs the extracted logistics features into the preset on-time delivery discrimination model. If the current order is an on-time delivery type order, it obtains customer features of the current order's customer. If the current order's customer is a target customer for the on-time guarantee service, it determines that the on-time guarantee service is recommended for the current order. Server 104 can provide a prompt message "Recommend on-time guarantee service" to terminal 102, which is a handheld terminal used by a salesperson (courier). Server 104 can also directly send relevant business data of the on-time guarantee service to the current order's customer terminal, so that the current order's customer can learn about the on-time guarantee service. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. Server 104 can be implemented as a standalone server or a server cluster consisting of multiple servers.

[0058] It can be understood that the logistics on-time guarantee service recommendation method of this application can also be directly applied to the terminal, which directly processes the current order data to determine whether to recommend the on-time guarantee service to the current order customer. The specific processing process is similar to the above and will not be repeated here.

[0059] In one embodiment, Figure 2 As shown, a logistics on-time guarantee service recommendation method is provided, which is applied to Figure 1 The following steps are used as an example to illustrate the server in the example:

[0060] S200: Obtain real-time order data of the current order.

[0061] The current order refers to the order currently being processed, which can be an order submitted to the server by a salesperson (courier) using a handheld terminal. Real-time order data for this order is obtained, which can include the logistics origin, destination, required transit stations (stations), terminal network, delivery date, logistics cargo weight, type, size, and special requirements in the order notes.

[0062] S400: Extracting logistics features from real-time order data, and inputting the extracted logistics features into a preset on-time delivery discrimination model.

[0063] Logistics feature data primarily includes the origin (shipping city), destination (delivery city), consignment type, destination outlet, product type, timeliness category, and delivery date. Specific logistics features can be selected based on historical verification data, prioritizing features that influence delivery timeliness. These extracted features are then fed into a pre-set on-time delivery discrimination model to determine whether the current order is eligible for on-time delivery. The pre-set on-time delivery discrimination model is a pre-built model that takes logistics features as input and outputs data indicating whether the order is eligible for on-time delivery or unavailable. This model can be trained based on historical order data. A classification threshold is set within the pre-set on-time delivery discrimination model. This threshold is set and adjusted based on historical order data, specifically, the on-time delivery achievement rate for historical orders. The pre-set on-time delivery discrimination model first estimates a specific coefficient and compares it with the classification threshold. If the coefficient is greater than the threshold, the current order is classified as eligible for on-time delivery; if it is less than the threshold, the current order is classified as unavailable for on-time delivery. Simply put, the above training process can be understood as training a parameterized inequality. The left side of the inequality is the logistics characteristics in the order data, and the right side of the inequality is the preset category classification threshold.

[0064] S600: If the current order is an on-time delivery type order, obtain the customer characteristics of the current order customer.

[0065] If the current order is an on-time delivery order, it indicates that there is a high probability that the current order can be delivered on time. If the on-time guarantee service is recommended to the user on this basis, on the one hand, the service promised to the user can basically be achieved. As long as there is a small probability of requiring overtime compensation, the on-time guarantee service will have a high profit. On the other hand, the service quality of the on-time guarantee service can also be guaranteed, which is conducive to maintaining the cooperative relationship with the current customer and guiding the current customer to promote this service. Based on the fact that the current order is an on-time delivery order, the customer characteristics of the current order customer are obtained. The customer characteristics can specifically include the customer's historical time-sensitive complaint rate, the customer's historical time-sensitive claim rate, and the customer's historical average number of visits to the time-sensitive route.

[0066] S800: When it is identified based on customer characteristics that the current order customer is a target customer for the on-time guarantee service, the on-time guarantee service is recommended.

[0067] If the customer characteristics extracted at S600 indicate that the ordering customer is a target customer for the on-time insurance service, then the on-time insurance service is recommended to the customer. This indicates that the current customer is highly likely to be interested in the on-time insurance service, and that the enterprise can generate revenue from the service. Specifically, the customer characteristics of the current ordering customer can be compared with the customer characteristics of customers who have previously purchased the on-time insurance service, and the similarity between the two can be compared. If the similarity between the two exceeds a preset similarity threshold, then the current ordering customer is a target customer. If not, then the current ordering customer is not a target customer.

[0068] The above-mentioned logistics on-time guarantee service recommendation method obtains the real-time order data of the current order, extracts logistics features from this real-time order data, and inputs these features into a preset on-time delivery identification model. If the current order is an on-time delivery type order, the customer features of the current order are obtained. If the current order customer is a target customer for the on-time guarantee service, the on-time guarantee service is recommended. This entire process, based on the preset on-time delivery identification model and the current order customer features, accurately identifies the target customers for the on-time guarantee service, avoids the inconvenience caused by large-scale, aimless promotions, and significantly improves the efficiency of on-time guarantee service recommendations.

[0069] like Figure 3 As shown, in one embodiment, before S400, the method further includes:

[0070] S320: Obtain order data of historical orders.

[0071] S340: Extract metadata features that affect delivery timeliness from order data of historical orders.

[0072] S360: The extracted metadata features are input into the initial logistic regression model for training to obtain a preset on-time delivery discrimination model. The category classification threshold of the initial logistic regression model is a preset probability threshold, which is related to the historical order timeliness achievement rate.

[0073] Historical order data can specifically include order data from the past month or the past three months. Specifically, historical orders from the same region, such as within the same province or municipality, can be collected. A pre-set on-time delivery discrimination model trained based on these historical orders can be used to recommend on-time delivery services to customers in that region. After obtaining the historical orders, metadata features influencing delivery timeliness are extracted from the order data in the historical list. These features primarily include the shipping city, receiving city, consignment type, destination outlet, product type, timeliness type, and delivery date. Furthermore, a historical on-time rate for shipping and receiving cities can be constructed. These metadata features influencing delivery timeliness are then fed into an initial logistic regression model for training, resulting in a pre-set on-time delivery discrimination model. Logistic regression (LR) models are used for modeling, which offer advantages over other algorithmic models, such as faster training speed and greater interpretability. The preset probability threshold is related to the historical order timeliness achievement rate, which can be manually modified based on the historical order timeliness achievement rate. If the model requires higher accuracy, the probability threshold can be increased based on the timeliness achievement rate. If you want to recommend on-time pricing services to more customers with a wider range of choices, you can lower the probability threshold based on the timeliness achievement rate. In actual applications, during the initial promotion, the real business scenarios have very high requirements for the accuracy of the model, so the threshold for model classification is adjusted to 0.8 (combined with the historical overall timeliness achievement rate of 0.85 for debugging). That is, when the prediction result is greater than 0.8, it is considered that the ticket can meet the on-time delivery requirement, otherwise it is considered not to meet the requirement, in order to ensure the accuracy of the model classification. The timeliness achievement rate is shown in Table 1 below.

[0074] Table 1 shows the timeliness achievement rate in a certain area over the past three months.

[0075]

[0076] In one embodiment, the extracted metadata features are input into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model including:

[0077] The extracted metadata features are input into multiple different initial logistic regression models for training to obtain multiple trained initial logistic regression models, wherein the number of iterations and regularization coefficients in different initial logistic regression models are different; preset sample test data are obtained, and the multiple trained initial logistic regression models are tested respectively based on the preset sample test data to obtain test results of the multiple trained initial logistic regression models; based on the test results and the preset model selection indicators, the optimal model is selected from the multiple trained initial logistic regression models as the preset on-time delivery discrimination model.

[0078] The training of the initial logistic regression model mainly involves adjusting the parameters in the logistic regression model so that its output value is closer to the true value, thereby achieving a more accurate prediction. In this embodiment, metadata features are selected to train multiple initial logistic regression models respectively. The number of iterations and regularization coefficients of these initial logistic regression models are different. Specifically, both parameters can be different, or one of the parameters can be different. After completing the training of multiple initial logistic regression models at the same time, multiple trained initial logistic regression models are obtained. These models are then tested using preset sample test data to obtain test results. Based on the test results and the preset model selection indicators, the optimal trained initial logistic regression model is selected as the preset on-time delivery discrimination model. Specifically, the preset sample test data is also data obtained based on historical orders. The sample data includes the metadata features of the order and the final result of whether it was delivered on time. The model selection indicators can specifically include three aspects: precision, recall, and AUC.

[0079] In actual applications, the training process for the preset on-time delivery discrimination model is as follows: the processed metadata features are placed into multiple logistic regression models. Different logistic regression models have different model parameters, such as the number of iterations (for optimality) and the regularization coefficient (to prevent overfitting). The results obtained under different parameters are compared, and the trained logistic regression model corresponding to the set of parameters that yields the best overall results is selected. The evaluation criteria for good results here must be based on the actual business situation. For example, in some scenarios, high accuracy is required, which requires ensuring a certain accuracy threshold and selecting a model with a high recall rate as the preset on-time delivery discrimination model. In one application example, the final selected metrics for the preset on-time delivery discrimination model are shown in Table 2.

[0080] Table 2 shows the indicators of the preset on-time arrival discrimination model

[0081] Accuracy Recall AUC 0.88 0.59 0.59

[0082] In one embodiment, inputting the extracted metadata features into an initial logistic regression model for training includes:

[0083] The chi-square test method is used to verify the correlation between the extracted metadata features and whether the order is delivered on time; metadata features with greater correlation are selected and input into the initial logistic regression model for training.

[0084] The extracted metadata features represent a large amount of data, containing a significant amount of feature data. To reduce the amount of data required for subsequent model training, metadata features with a high correlation with on-time order delivery can be selected. In this embodiment, a chi-square test is used to verify the correlation between the extracted metadata features and on-time order delivery. These highly correlated metadata features are then selected and input into the initial logistic regression model for training. This significantly reduces the amount of data required for subsequent model training. Furthermore, before inputting these highly correlated metadata features into the initial logistic regression model for training, these highly correlated metadata features can be normalized using mean coding to obtain normalized features. These normalized features are then input into the initial logistic regression model for training.

[0085] Furthermore, since the metadata features with high correlation still contain a large number of high-cardinality qualitative features, using dummy coding to convert these features will result in the appearance of a large number of sparse matrices. For these reasons, average coding can be further used to process these high-cardinality qualitative features. This encoding method is a supervised encoding method that takes a weighted average of the prior probability and the posterior probability, and then converts the type variable into a probability value. The corresponding probability value will serve as the input feature of the model, that is, a high-dimensional category value is converted into a probability value. Compared with one-hot, the feature dimension will be greatly reduced. The detailed calculation formula is as follows:

[0086] P mean =λ*prior+(1-λ)*posterior

[0087] In the formula, λ is the weight; prior is the prior probability; posterior is the posterior probability, P mean is the weighted average of the prior and posterior probabilities. Here, using the shipping city as an example, prior is the overall average probability of positive samples over a rolling 30-day period, and posterior is the average probability of positive samples accumulated for each city over the rolling 30-day period. λ is determined by the sample distribution based on the number of features enumerated and analyzed.

[0088] like Figure 4 As shown, in one embodiment, the above-mentioned logistics on-time guarantee service recommendation method further includes:

[0089] S720: Obtain seed customers who purchased the on-time insurance service in the historical records;

[0090] S740: extracting customer characteristics of seed customers;

[0091] S760: Determine whether the current order customer is a target customer for the on-time guarantee service based on the customer characteristics of the current order customer and the customer characteristics of the seed customer.

[0092] In the historical records, there are still relatively few customers who have received on-time insurance. In this embodiment, the idea of ​​look-alike is introduced, and existing seed customers are used for expansion to mine new target customers for recommendation. Customer characteristics mainly include the customer's historical on-time insurance ratio, customer's historical case volume, customer's historical timeliness complaint rate, customer's historical timeliness claim rate, customer's historical ticket average number of visits to timeliness routing, etc. Figure 5 As shown, first obtain the seed customers who purchased the on-time insurance service in the historical records, extract the customer characteristics of the seed customers, and judge whether the current order customer is the target customer of the on-time insurance service based on the customer characteristics of the current order customer and the customer characteristics of the seed customers.

[0093] Furthermore, the above extension scheme can be constructed into a model-based extension scheme in actual application. That is, a discriminant model is constructed to determine whether it is an on-time guarantee business. During the model construction process, the seed user is marked as a positive sample, and then the current order customer is represented by a vector. The similarity score between the candidate user and the seed customer in the positive sample is calculated through cosine similarity. According to business needs, by analyzing the data of the recent period, the similarity score threshold is set to 0.8, that is, customers with a similarity score of 0.8 or above with the seed customer are recommended customers. The goal of this model is to select customers that can be recommended. The detailed calculation process and formula are as follows:

[0094] 1. Calculate the average value of each feature of the seed user:

[0095]

[0096] FC=(F1C,F2C,...,F n C)

[0097] Where N is the number of seed samples, j is the specified feature, and n is the number of customer features.

[0098] 2. Calculate the cosine similarity score between each customer and the average value of the seed customers:

[0099] Denote the customer as P k =(f1, f2, ..., f n ), where historical customers use their actual f n (n=1,2,...) value, the new customer defaults to f n (n=1, 2, ...) are all 0. The cosine similarity is calculated as:

[0100]

[0101] in:

[0102]

[0103] Furthermore, as new target customers are constantly added as seed customers, the positive sample seed customer set can be updated. Specifically, based on actual business scenarios, new seed customers (target customers) are constantly generated over time. Therefore, new seed customers are screened from daily offline data and added to the training dataset, and the model is retrained. This dynamic daily model update method can more effectively ensure the number and accuracy of new seed customers.

[0104] In practical applications, the technical principle architecture of the logistics on-time guarantee service recommendation method of this application is as follows Figure 6 As shown, the overall concept is to realize the recommendation of logistics on-time insurance service by constructing a preset on-time delivery discrimination model and a customer-whether-customer-purchase-on-time insurance discrimination model. The preset on-time delivery discrimination model determines whether the current order can be delivered on time. If it can be delivered on time, the customer-whether-customer-purchase-on-time insurance discrimination model is used to determine whether the customer will purchase the on-time insurance service. If it is determined that the customer will purchase the on-time insurance service based on the customer characteristics, the on-time insurance service is recommended to the customer.

[0105] In practical application, the logistics on-time guarantee service recommendation method of this application has the following significant technical effects and social benefits:

[0106] 1. The prediction model for order timeliness can effectively filter and screen order timeliness, screening approximately five million shipping orders per day;

[0107] 2. Introducing the lookalike method can identify a large number of potential target customers similar to seed users, increase the probability of recommended purchases, and promote business growth;

[0108] 3. Update seed customer offline data in real time every day and train new models to optimize model effects and expand the scope of seed customer screening;

[0109] 4. By organically combining the two, the company not only improves user experience but also effectively promotes the company's value-added services and promotes the company's revenue generation.

[0110] It should be understood that, although the steps in the above-mentioned flowcharts are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the above-mentioned flowcharts may include multiple steps or multiple stages, and these steps or stages are not necessarily performed at the same time, but can be performed at different times. The execution order of these steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0111] like Figure 7 As shown, the present application also provides a logistics on-time guarantee service recommendation device, the device comprising:

[0112] The order acquisition module 200 is used to obtain the real-time order data of the current order;

[0113] Logistics feature extraction module 400, for extracting logistics features from real-time order data and inputting the extracted logistics features into a preset on-time delivery discrimination model;

[0114] The customer feature extraction module 600 is used to obtain the customer features of the current order customer when the current order is an on-time delivery type order;

[0115] The service recommendation module 800 is used to recommend the on-time guarantee service when it is identified that the current order customer is a target customer of the on-time guarantee service based on customer characteristics.

[0116] The aforementioned on-time logistics guarantee service recommendation device obtains real-time order data for the current order, extracts logistics features from this real-time order data, and inputs these features into a pre-set on-time delivery discrimination model. If the current order is an on-time delivery order, the device then obtains the customer features of the current order's customer. If the current order's customer is a target customer for the on-time guarantee service, the on-time guarantee service is recommended. This dual process, based on the pre-set on-time delivery discrimination model and the current order's customer features, accurately identifies target customers for the on-time guarantee service, avoiding the inconvenience caused to customers by widespread, untargeted promotions and significantly improving the efficiency of on-time guarantee service recommendations.

[0117] In one embodiment, the above-mentioned logistics on-time delivery service recommendation module also includes a model building module for obtaining order data of historical orders; extracting metadata features that affect delivery timeliness from the order data of historical orders; inputting the extracted metadata features into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model, and the category division threshold of the initial logistic regression model is a preset probability threshold, which is related to the historical order timeliness achievement rate.

[0118] In one embodiment, the model building module is further used to input the extracted metadata features into multiple different initial logistic regression models for training, thereby obtaining multiple trained initial logistic regression models, wherein the number of iterations and regularization coefficients in different initial logistic regression models are different; obtaining preset sample test data, and testing the multiple trained initial logistic regression models respectively according to the preset sample test data to obtain test results of the multiple trained initial logistic regression models; and selecting the optimal model from the multiple trained initial logistic regression models as the preset on-time delivery discrimination model based on the test results and the preset model selection indicators.

[0119] In one embodiment, the model building module is further used to verify the correlation between the extracted metadata features and whether the order is delivered on time using a chi-square test method; metadata features with greater correlation are selected and input into the initial logistic regression model for training.

[0120] In one embodiment, the model building module is further configured to normalize metadata features with high correlation using mean coding to obtain normalized features; and input the normalized features into the initial logistic regression model for training.

[0121] In one embodiment, the above-mentioned logistics on-time guarantee service recommendation device also includes a target customer identification module, which is used to obtain seed customers who purchased the on-time guarantee service in the historical records; extract the customer characteristics of the seed customers; and determine whether the current order customer is the target customer of the on-time guarantee service based on the customer characteristics of the current order customer and the customer characteristics of the seed customer.

[0122] In one embodiment, the target customer identification module is further configured to calculate the cosine similarity between the current order customer and the seed customer; and determine whether the current order customer is a target customer for the on-time guarantee service based on the cosine similarity.

[0123] The specific definitions of the on-time logistics guarantee service recommendation device can be found in the definitions of the on-time logistics guarantee service recommendation method described above and will not be further elaborated here. Each module in the above-mentioned on-time logistics guarantee service recommendation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of these modules can be embedded in or independent of a processor in a computer device in hardware form, or stored in a computer device memory in software form, so that the processor can call and execute the corresponding operations of each module.

[0124] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 8As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store data such as historical order data and a preset on-time delivery discrimination model. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for recommending logistics on-time delivery services is implemented.

[0125] Those skilled in the art will understand that Figure 8 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0126] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0127] Get real-time order data for the current order;

[0128] Extract logistics features from real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model;

[0129] If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer;

[0130] When the current order customer is identified as a target customer for the on-time guarantee service based on customer characteristics, the on-time guarantee service is recommended.

[0131] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0132] Obtain order data for historical orders; extract metadata features that affect delivery timeliness from the order data for historical orders; input the extracted metadata features into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model. The category division threshold of the initial logistic regression model is a preset probability threshold, which is related to the historical order timeliness achievement rate.

[0133] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0134] The extracted metadata features are input into multiple different initial logistic regression models for training to obtain multiple trained initial logistic regression models, wherein the number of iterations and regularization coefficients in different initial logistic regression models are different; preset sample test data are obtained, and the multiple trained initial logistic regression models are tested respectively based on the preset sample test data to obtain test results of the multiple trained initial logistic regression models; based on the test results and the preset model selection indicators, the optimal model is selected from the multiple trained initial logistic regression models as the preset on-time delivery discrimination model.

[0135] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0136] The chi-square test method is used to verify the correlation between the extracted metadata features and whether the order is delivered on time; metadata features with greater correlation are selected and input into the initial logistic regression model for training.

[0137] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0138] The metadata features with high correlation are normalized by using mean coding to obtain normalized features; the normalized features are input into the initial logistic regression model for training.

[0139] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0140] Obtain seed customers who purchased the on-time insurance service from historical records; extract the customer characteristics of the seed customers; and determine whether the current order customer is a target customer for the on-time insurance service based on the customer characteristics of the current order customer and the customer characteristics of the seed customers.

[0141] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0142] Calculate the cosine similarity between the current order customer and the seed customer; based on the cosine similarity, determine whether the current order customer is the target customer of the on-time guarantee service.

[0143] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0144] Get real-time order data for the current order;

[0145] Extract logistics features from real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model;

[0146] If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer;

[0147] When the current order customer is identified as a target customer for the on-time guarantee service based on customer characteristics, the on-time guarantee service is recommended.

[0148] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0149] Obtain order data for historical orders; extract metadata features that affect delivery timeliness from the order data for historical orders; input the extracted metadata features into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model. The category division threshold of the initial logistic regression model is a preset probability threshold, which is related to the historical order timeliness achievement rate.

[0150] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0151] The extracted metadata features are input into multiple different initial logistic regression models for training to obtain multiple trained initial logistic regression models, wherein the number of iterations and regularization coefficients in different initial logistic regression models are different; preset sample test data are obtained, and the multiple trained initial logistic regression models are tested respectively based on the preset sample test data to obtain test results of the multiple trained initial logistic regression models; based on the test results and the preset model selection indicators, the optimal model is selected from the multiple trained initial logistic regression models as the preset on-time delivery discrimination model.

[0152] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0153] The chi-square test method is used to verify the correlation between the extracted metadata features and whether the order is delivered on time; metadata features with greater correlation are selected and input into the initial logistic regression model for training.

[0154] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0155] The metadata features with high correlation are normalized by using mean coding to obtain normalized features; the normalized features are input into the initial logistic regression model for training.

[0156] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0157] Obtain seed customers who purchased the on-time insurance service from historical records; extract the customer characteristics of the seed customers; and determine whether the current order customer is a target customer for the on-time insurance service based on the customer characteristics of the current order customer and the customer characteristics of the seed customers.

[0158] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0159] Calculate the cosine similarity between the current order customer and the seed customer; based on the cosine similarity, determine whether the current order customer is the target customer of the on-time guarantee service.

[0160] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0161] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0162] The above embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for recommending on-time logistics services, characterized in that: The method comprises: Obtain real-time order data for the current order; the current order refers to the order currently being processed, which is the order reported to the server by the salesperson through his handheld terminal; Extracting logistics features from the real-time order data and inputting the extracted logistics features into a preset on-time delivery discrimination model; the preset on-time delivery discrimination model is used to output whether the current order is an on-time delivery type order or an undeliverable type order; the logistics features are features that affect delivery timeliness, including the starting point, end point, consignment type, destination outlet, product type, timeliness type, and delivery date; the on-time delivery discrimination model is trained based on historical order data, and a classification threshold is set in the on-time delivery discrimination model; If the current order is an on-time delivery order, obtain the customer characteristics of the current order customer; Compare the customer characteristics of the current order customer with the average customer characteristics of multiple seed customers who have purchased the on-time insurance service in the past to determine the similarity between the two. If the similarity is greater than a preset similarity threshold, it indicates that the current order customer is a target customer. If not, it indicates that the current order customer is not a target customer. When it is identified according to the customer characteristics that the current order customer is a target customer of the on-time guarantee service, the on-time guarantee service is recommended.

2. The method according to claim 1, characterized in that The method further includes: extracting logistics features from the real-time order data and inputting the extracted logistics features into a preset on-time delivery discrimination model; Get order data of historical orders; Extracting metadata features that affect delivery timeliness from the order data of the historical orders; The extracted metadata features are input into an initial logistic regression model for training to obtain a preset on-time delivery discrimination model. The category division threshold of the initial logistic regression model is a preset probability threshold, and the preset probability threshold is related to the historical order timeliness achievement rate.

3. The method according to claim 2, characterized in that The step of inputting the extracted metadata features into the initial logistic regression model for training to obtain the preset on-time delivery discrimination model includes: Inputting the extracted metadata features into a plurality of different initial logistic regression models for training to obtain a plurality of trained initial logistic regression models, wherein the number of iterations and regularization coefficients in the different initial logistic regression models are different; Acquire preset sample test data, and test the multiple trained initial logistic regression models respectively according to the preset sample test data to obtain test results of the multiple trained initial logistic regression models; According to the test results and the preset model selection index, the optimal model is selected from the multiple trained initial logistic regression models as the preset on-time delivery discrimination model.

4. The method according to claim 2, characterized in that Inputting the extracted metadata features into the initial logistic regression model for training includes: The chi-square test method is used to verify the correlation between the extracted metadata features and whether the order is delivered on time; Metadata features with greater correlation are selected and input into the initial logistic regression model for training.

5. The method according to claim 4, characterized in that The step of selecting metadata features with greater correlation and inputting them into the initial logistic regression model for training includes: The average coding method is used to normalize the metadata features with large correlation to obtain the normalized features; The normalized features are input into the initial logistic regression model for training.

6. The method according to claim 1, wherein Also includes: Obtain seed customers who have purchased the on-time insurance service in the historical records; Extracting customer characteristics of the seed customers; According to the customer characteristics of the current order customer and the customer characteristics of the seed customer, it is determined whether the current order customer is a target customer of the on-time guarantee service.

7. The method according to claim 6, characterized in that The determining, based on the customer characteristics of the current order customer and the customer characteristics of the seed customer, whether the current order customer is a target customer for the on-time guarantee service includes: Calculating the cosine similarity between the current order customer and the seed customer; According to the cosine similarity, it is determined whether the current order customer is a target customer of the on-time guarantee service.

8. A logistics on-time guarantee service recommendation device, characterized in that: The device comprises: The order acquisition module is used to obtain real-time order data of the current order; the current order refers to the order currently being processed, which is the order reported to the server by the salesperson through his handheld terminal; A logistics feature extraction module is configured to extract logistics features from the real-time order data and input the extracted logistics features into a preset on-time delivery discrimination model; the preset on-time delivery discrimination model is configured to output whether the current order is an on-time delivery type order or an undeliverable type order; the logistics features are features that affect delivery timeliness, including the starting point, end point, consignment type, destination outlet, product type, timeliness type, and delivery date; the on-time delivery discrimination model is trained based on historical order data and is configured with a classification threshold; A customer feature extraction module, configured to obtain customer features of the current order customer when the current order is an on-time delivery type order; a target customer identification module, configured to compare the customer characteristics of the current order customer with the average customer characteristics of multiple seed customers who have historically purchased the on-time insurance service, and determine the similarity between the two; if the similarity is greater than a preset similarity threshold, it indicates that the current order customer is a target customer; if not, it indicates that the current order customer is not a target customer; The service recommendation module is used to recommend the on-time guarantee service when it is identified that the current order customer is a target customer of the on-time guarantee service according to the customer characteristics.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images