Prediction Method, Device, Storage Medium and Electronic Device for Customer Purchase Preferences

Abstract

The present application discloses a method, apparatus, storage medium and electronic device for predicting customer purchase preferences. It relates to the field of financial technology. The method includes: determining the purchase probabilities of a customer for purchasing multiple products to obtain multiple purchase probabilities; taking the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; determining multiple preference types of a target customer in the case where the purchase characteristic data corresponding to the first purchase type changes; calculating the preference characteristic data corresponding to the multiple preference types through a target algorithm model to obtain a calculation result; predicting the target preference type of the target customer according to the calculation result. Through the present application, the problem that the prediction of customer purchase preferences in the related art is not accurate enough is solved.

Description

Technical Field

[0001] The present application relates to the field of fintech, and in particular, to a method, device, storage medium and electronic device for predicting customer purchase preferences. Background Art

[0002] In the related art, the customer purchase preference tendency model mainly predicts based on the existing characteristic data of customers, and has the following defects: First, there is a model black box phenomenon in the deep learning model. That is, the model only outputs the result, and it cannot intuitively tell the user which features have the greatest impact on the result, whether the impact of the features on the result is positive or negative, so it is not convenient for users to carry out marketing for customers in actual use, and it is impossible to tell the customer why a certain type of product is promoted; Second, scenario analysis cannot be implemented. For example, analyze what kind of transformation of the customer's characteristics will lead to a change in customer preferences. Some methods carry out scenario analysis by changing the feature variable values to change the prediction results; however, these changes may be artificial assumptions and do not match the actual training data. Therefore, the problem of inaccurate prediction of customer purchase preferences is caused.

[0003] In view of the problem of inaccurate prediction of customer purchase preferences in the related art, no effective solution has been proposed yet. Summary of the Invention

[0004] The main purpose of the present application is to provide a method, device, storage medium and electronic device for predicting customer purchase preferences, so as to solve the problem of inaccurate prediction of customer purchase preferences in the related art.

[0005] To achieve the above object, according to one aspect of the present application, a method for predicting customer purchase preferences is provided. The method includes: determining the purchase probabilities of a customer for purchasing multiple products to obtain multiple purchase probabilities; taking the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; when the purchase characteristic data corresponding to the first purchase type changes, determining multiple preference types of the target customer, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchase times by the customer; calculating the preference characteristic data corresponding to the multiple preference types through a target algorithm model to obtain a calculation result, where the product corresponding to each preference characteristic data is purchased by the target customer less times than the product corresponding to the purchase characteristic data; predicting the target preference type of the target customer according to the calculation result, where the product corresponding to the target preference type is purchased by the target customer with a purchase probability lower than that of the product corresponding to the first purchase type and higher than that of the products corresponding to other types purchased by the target customer.

[0006] Further, determine the purchase probabilities of a target customer for purchasing multiple products, and obtain multiple purchase probabilities, including: collecting purchase characteristic data of different products; inputting each purchase characteristic data into a target algorithm model to calculate the purchase probability corresponding to each product, and obtaining multiple purchase probabilities.

[0007] Further, calculate the preference characteristic data corresponding to multiple preference types through a target algorithm model, and obtain calculation results, including: determining the characteristic values in each preference characteristic data; calculating a first mean value according to the multiple characteristic values, where the first mean value is used to represent the average value of the multiple characteristic values; taking the first mean value as one of the calculation results.

[0008] Further, before predicting the target preference type of the target customer according to the calculation results, the method further includes: subtracting each characteristic value from the first mean value to obtain multiple first differences; taking the minimum value among the multiple first differences as the first target difference.

[0009] Further, calculate the preference characteristic data corresponding to multiple preference types through a target algorithm model, and obtain calculation results, including: inputting the multiple preference characteristic data into the target algorithm model to output multiple target preference characteristic data; determining the target characteristic values in each target preference characteristic data; calculating a second mean value according to the multiple target characteristic values, where the second mean value is used to represent the average value of the multiple target characteristic values; taking the second mean value as one of the calculation results.

[0010] Further, before predicting the target preference type of the target customer according to the calculation results, the method further includes: subtracting the characteristic value in each target preference characteristic data from the second mean value to obtain multiple second differences; taking the minimum value among the multiple second differences as the second target difference; determining a target probability according to the first target difference and the second target difference; determining the target preference type according to the target probability.

[0011] Further, determining the target probability according to the first target difference and the second target difference includes: performing a priority ranking on the multiple preference types through the first target difference and the second target difference to obtain the ranked preference types; determining a target matrix according to the ranked preference types; calculating the target probability by calculating the target matrix.

[0012] To achieve the above object, according to another aspect of the present application, there is provided a prediction device for customer purchase preferences. The device includes: a first determination unit for determining the purchase probabilities of a customer purchasing multiple products to obtain multiple purchase probabilities; a second determination unit for taking the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; a third determination unit for determining multiple preference types of a target customer when the purchase characteristic data corresponding to the first purchase type changes, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchase times by the customer; a first calculation unit for calculating, through a target algorithm model, the preference characteristic data corresponding to the multiple preference types to obtain a calculation result, where the product corresponding to each preference characteristic data is purchased by the target customer less times than the product corresponding to the purchase characteristic data; and a prediction unit for predicting the target preference type of the target customer according to the calculation result, where the product corresponding to the target preference type is purchased by the target customer with a purchase probability lower than that of the product corresponding to the first purchase type and higher than the purchase probability of the product corresponding to other types purchased by the target customer.

[0013] Further, the first determination unit includes: a collection module for collecting the purchase characteristic data of different products; and a first calculation module for inputting each purchase characteristic data into the target algorithm model to calculate the purchase probability corresponding to each product to obtain multiple purchase probabilities.

[0014] Further, the first calculation unit includes: a first determination module for determining the feature values in each preference characteristic data; a second calculation module for calculating a first mean value according to the multiple feature values, where the first mean value is used to represent the average value of the multiple feature values; and a second determination module for taking the first mean value as one of the calculation results.

[0015] Further, the device further includes: a second calculation unit for subtracting each feature value from the first mean value to obtain multiple first differences before predicting the target preference type of the target customer according to the calculation result; and a fourth determination unit for taking the minimum value among the multiple first differences as the first target difference.

[0016] Further, the first calculation unit includes: an output module for inputting the multiple preference characteristic data into the target algorithm model to output multiple target preference characteristic data; a third determination module for determining the target feature values in each target preference characteristic data; a third calculation module for calculating a second mean value according to the multiple target feature values, where the second mean value is used to represent the average value of the multiple target feature values; and a fourth determination module for taking the second mean value as one of the calculation results.

[0017] Further, the device further includes: a third calculation unit, configured to subtract the feature value in each target preference feature data from the second mean value to obtain a plurality of second differences before predicting the target preference type of the target customer according to the calculation result; a fifth determination unit, configured to use the minimum value among the plurality of second differences as the second target difference; a sixth determination unit, configured to determine a target probability according to the first target difference and the second target difference; and a seventh determination unit, configured to determine the target preference type according to the target probability.

[0018] Further, the sixth determination unit includes: a sorting module, configured to perform priority sorting on a plurality of preference types through the first target difference and the second target difference to obtain the sorted preference types; a fifth determination module, configured to determine a target matrix according to the sorted preference types; and a fourth calculation module, configured to obtain a target probability by calculating the target matrix.

[0019] By means of the present application, the following steps are adopted: determining the purchase probabilities of a customer for purchasing a plurality of products to obtain a plurality of purchase probabilities; using the type of the product with the highest purchase probability among the plurality of purchase probabilities as the first purchase type; determining a plurality of preference types of a target customer in the case where the purchase feature data corresponding to the first purchase type changes, where the purchase feature data is the feature data corresponding to the product with the most purchase times by the customer; calculating, through a target algorithm model, the preference feature data corresponding to the plurality of preference types to obtain a calculation result, where the product corresponding to each preference feature data is purchased by the target customer fewer times than the product corresponding to the purchase feature data; and predicting the target preference type of the target customer according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than the purchase probability of the product corresponding to the first purchase type and higher than the purchase probability of the target customer purchasing other types of products, thereby solving the problem in the related art that the prediction of the customer's purchase preference is not accurate enough. By calculating, through the target algorithm model, the preference feature data corresponding to the plurality of preference types and predicting the target preference type of the target customer according to the calculation result, the effect of improving the prediction accuracy of the customer's purchase preference is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The schematic embodiments of the present application and the descriptions thereof are used to explain the present application and do not constitute an improper limitation to the present application. In the drawings:

[0021] Figure 1 is a flowchart of a method for predicting a customer's purchase preference according to an embodiment of the present application;

[0022] Figure 2 is an overall operation schematic diagram of a method for predicting a customer's purchase preference according to an embodiment of the present application;

[0023] Figure 3 It is a schematic diagram for predicting and analyzing changes in customer purchase preferences of the customer purchase preference prediction method provided by an embodiment of the present application;

[0024] Figure 4 It is a schematic diagram of the customer purchase preference prediction device provided by an embodiment of the present application;

[0025] Figure 5 It is a schematic diagram of the network architecture of the customer purchase preference prediction electronic device provided by an embodiment of the present application. Detailed implementation manners

[0026] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments may be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0027] In order to enable those skilled in the art to better understand the solution of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above drawings are used to distinguish similar objects, and do not necessarily need to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to implement the embodiments of the present application described herein. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0029] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in the present disclosure are all information and data that have been authorized by the user or fully authorized by all parties.

[0030] The present invention will be described below in combination with the preferred implementation steps. Figure 1 It is a flowchart of the customer purchase preference prediction method provided by an embodiment of the present application. As Figure 1 shown, the method includes the following steps:

[0031] Step S101, determine the purchase probabilities of a customer purchasing multiple products, and obtain multiple purchase probabilities.

[0032] Step S102, use the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type.

[0033] For example, determine 50 customers such as Customer A, Customer B, Customer C, etc., and the purchase probabilities of purchasing a 50-square-meter house (corresponding to different products in this application), a 70-square-meter house, and a 90-square-meter house, and obtain the probability magnitudes of different customers purchasing different house areas.

[0034] Optionally, in the method for predicting customer purchase preferences provided in the embodiments of this application, determining the purchase probabilities of a target customer purchasing multiple products and obtaining multiple purchase probabilities includes: collecting purchase characteristic data of different products; inputting each purchase characteristic data into a target algorithm model to calculate the purchase probability corresponding to each product, and obtaining multiple purchase probabilities.

[0035] For example, collect the purchase data of Customer A purchasing a house, the purchase data of Customer B purchasing a house, the purchase data of Customer C purchasing a house,... the purchase data of 50 customers purchasing a house, and input each purchase data into the original deep learning model f(x). For example, multiple purchase data uses a set of feature variables [x 1 , x 2 , …, x n to represent. Input each feature variable into the f(x) model, and the obtained purchase probability corresponding to each product is t 1 , t 2 ,..., t K . This application calculates the purchase probabilities of different features through the f(x) learning model, and can further more accurately determine the product type with the highest purchase probability of the customer group, that is, the most popular product type of the customer group.

[0036] For example, according to the above example, it is obtained that the purchase probability corresponding to purchasing a 70-square-meter house is the largest, and the number of customers is 30. The product of the 70-square-meter house type is used as the first purchase type in this application.

[0037] Step S103, when the purchase characteristic data corresponding to the first purchase type changes, determine multiple preference types of the target customer, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchase times by the customer.

[0038] For example, if the product of the housing type of 70 square meters is used as the first purchase type in this application, and the housing area has a new addition of 110 square meters among the previous house types of 50, 70, and 90 square meters, determine the preference types for purchasing other house types among the 30 customers corresponding to the first purchase type when the house type changes.

[0039] Step S104, calculate the calculation result through the target algorithm model for the preference feature data corresponding to multiple preference types, where the number of times the product corresponding to each preference feature data is purchased by the target customer is lower than the product corresponding to the purchase feature data.

[0040] For example, assume that the first purchase type of this customer is k, and determine a group where δ is the characteristic variable when there is a slight change in type k, such that is the characteristic value in multiple preference feature data, and the value distribution conforms to the real data distribution.

[0041] Optionally, in the customer purchase preference prediction method provided in the embodiments of this application, calculating the calculation result through the target algorithm model for the preference feature data corresponding to multiple preference types includes: determining the characteristic values in each preference feature data; calculating the first mean according to the multiple characteristic values, where the first mean is used to represent the average value of the multiple characteristic values; and taking the first mean as one of the calculation results.

[0042] For example, through calculate the first mean, where class i represents the mean value of the characteristic values in each preference feature data corresponding to the customer group of this class under the i-th classification, and this value is the mean value of the real values, and M i represents that there are M i customers (corresponding to the target customers in this application) under the i-th classification. The accurate prediction of the preference type is further improved by calculating the first mean.

[0043] Optionally, in the customer purchase preference prediction method provided in the embodiments of this application, calculating the calculation result through the target algorithm model for the preference feature data corresponding to multiple preference types includes: inputting the multiple preference feature data into the target algorithm model, and outputting multiple target preference feature data; determining the target characteristic values in each target preference feature data; calculating the second mean according to the multiple target characteristic values, where the second mean is used to represent the average value of the multiple target characteristic values; and taking the second mean as one of the calculation results.

[0044] For example, through calculate the second mean, where proto iIt represents the average of multiple target preference feature data obtained from the output layer of the last layer of the f(x) model corresponding to the target customer group of this category under the i-th category. Here, outputlayer(·) represents the output layer of the last layer of the deep neural network f(x) of the purchase preference model, and this value is the average of the output layer of the model, M i It represents that there are M i customers under the i-th category (corresponding to the target customers in this application). By calculating the second average, the accurate prediction of the preference type is further improved.

[0045] It should be noted that introducing the formula L dist =α||δ k || 2 means that the change amount of the feature variable value is as small as possible, so that the feature value should be relatively close to the true value after the change. Here, α is a hyperparameter set by humans.

[0046] Step S105, predicting the target preference type of the target customer according to the calculation result. Among them, the purchase probability of the product corresponding to the target preference type is lower than that of the product corresponding to the first purchase type, and higher than that of the products corresponding to other types purchased by the target customer.

[0047] Specifically, before predicting the target preference type of the target customer according to the calculation result, through the formula L dist =α||δ k || 2 calculate the minimum value of the change amount of the feature variable value, so that the feature value should be relatively close to the true value after the change. Here, α is a hyperparameter set by humans. Through the formula γ represents the upper limit of the maximum distance between the two, which is a hyperparameter set by humans. Calculate the difference between and . represents the probability that f(x) is predicted to be the k-th category (the k-th category is the category with the highest probability among all categories), represents the probability that f(x) is predicted to be a non-k category, which represents the upper limit of the maximum distance between the two, and is a hyperparameter set by humans. That is, in this application, L dist and L pred can also be used as one of the calculation results to further determine which category is the closest to the probability corresponding to the k-th category when the customer's preference changes from the k-th category to a non-k category.

[0048] Optionally, in the customer purchase preference prediction method provided in the embodiments of this application, before predicting the target preference type of the target customer according to the calculation result, the method further includes: subtracting each feature value from the first average to obtain multiple first differences; taking the minimum value of the multiple first differences as the first target difference.

[0049] Specifically, through the formula L class =β||(x k +δ k ) - class j || 2 Subtract each eigenvalue from the first mean to obtain multiple first differences, and select the smallest first difference as the first target difference. β is a hyperparameter set by humans, and the calculation of the first difference further improves the accurate prediction of the preference type.

[0050] Optionally, in the prediction method of customer purchase preference provided in the embodiments of the present application, before predicting the target preference type of the target customer according to the calculation result, the method further includes: subtracting the eigenvalue in each target preference feature data from the second mean to obtain multiple second differences; taking the minimum value of the multiple second differences as the second target difference; determining the target probability according to the first target difference and the second target difference; and determining the target preference type according to the target probability.

[0051] Specifically, through the formula L proto =λ||outputlayer(x k +δ k ) - proto j || 2 Subtract the eigenvalue in each target preference feature data from the second mean to obtain multiple second means, and select the smallest second difference as the second target difference. That is, the present application further improves the accurate prediction of the target probability corresponding to the preference type through calculation

[0052] Optionally, in the prediction method of customer purchase preference provided in the embodiments of the present application, determining the target probability according to the first target difference and the second target difference includes: prioritizing multiple preference types through the first target difference and the second target difference to obtain the sorted preference types; determining the target matrix according to the sorted preference types; and calculating the target probability through the target matrix.

[0053] For example, as Figure 2 shown, the object of analysis is customer Φ, the purchase preference of the customer is k, the feature variable used is x k , and the goal is to find δ * such that the purchase preference of the customer changes from k to other categories, and the value of the changed x k +δ * conforms to the distribution of the real training data. Let I i =||outputlayer(x k ) - proto i ||​2 +||(x k )-class i || 2 , i ≠ k, sort in ascending order of the values, assumed to be I 1 ≤ I 2 ≤... ≤ I K-1 , take I 1 , find obtain successively take I 1 , I 2 ,..., take at least 1 class and at most 5 classes, and solve them separately. Assume that we obtain As Figure 3 shown, establish a change prediction model. Taking as an example corresponding preference type is κ 1 , corresponding preference type is κ 2 , corresponding to preference κ 3 , corresponding to preference κ 4 , corresponding to preference κ 5 , and calculate the customer group preference migration matrix according to multiple preference types. Assume that the preferences after the change of preference k all fall within κ 1 , κ 2 , κ 3 , κ 4 , κ 5 Among the five types of preferences. Assume that there are 3 customers with preference k. Let: The 5 - type preference ranking of customer 1 is κ 1 , κ 2 , κ 3 , κ 4 , κ 5 , the 5 - type preference ranking of customer 2 is κ 1 , κ 3 , κ 2 , κ 5 , κ 4 , the 5 - type preference ranking of customer 3 is κ 3 , κ 1 , κ 5 , κ 2 , κ 4 , κ 1 , κ 2 , κ 3 , κ 4 5

[0054] κ 4 = 0, κ5 = 0. The second - ranked preference, κ 1 , κ 2 , κ 3 , κ 4 , κ 5 The probabilities are as follows: κ 4 = 0, κ 5 = 0.

[0055] By analogy, the preference type with the highest occurrence probability at each rank can be calculated as the target preference type in this application.

[0056] It should be noted that this application can also establish a simple linear equation (any other form of model is acceptable) K = θ 0 + θ 1 ·δ 1 + θ 2 ·δ 2 +,...,+ θ n ·δ n , where the input feature variables are δ 1 , δ 2 ,..., δ n , and the corresponding feature values The specific values, and the target variable is the corresponding preference κ 1 , κ 2 , κ 3 , κ 4 , κ 5 . Solve for [θ 0 , θ 1 , θ 2 ,..., θ n . Each coefficient represents the degree of influence of different feature changes on preference changes.

[0057] In summary, the prediction method for customer purchase preferences provided by the embodiments of the present application determines the purchase probabilities of a customer purchasing multiple products to obtain multiple purchase probabilities, takes the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type, and determines multiple preference types of the target customer when the purchase characteristic data corresponding to the first purchase type changes, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchase times by the customer. The calculation is performed on the preference characteristic data corresponding to the multiple preference types through the target algorithm model to obtain a calculation result, where the number of times the product corresponding to each preference characteristic data is purchased by the target customer is lower than the number of times the product corresponding to the purchase characteristic data is purchased. The target preference type of the target customer is predicted according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than the purchase probability of the product corresponding to the first purchase type and higher than the purchase probability of the products corresponding to other types purchased by the target customer, solving the problem of inaccurate prediction of customer purchase preferences in the related art. The calculation is performed on the preference characteristic data corresponding to the multiple preference types through the target algorithm model, and the target preference type of the target customer is predicted according to the calculation result, thereby achieving the effect of improving the prediction accuracy of customer purchase preferences.

[0058] It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in a different order than here.

[0059] The embodiments of the present application also provide a prediction device for customer purchase preferences. It should be noted that the prediction device for customer purchase preferences in the embodiments of the present application can be used to execute the prediction method for customer purchase preferences provided by the embodiments of the present application. The following introduces the prediction device for customer purchase preferences provided by the embodiments of the present application.

[0060] Figure 4 is a schematic diagram of the prediction device for customer purchase preferences according to the embodiments of the present application. As Figure 4 shown, the device includes: a first determination unit 401, a second determination unit 402, a third determination unit 403, a first calculation unit 404, and a prediction unit 405.

[0061] Specifically, the first determination unit 401 is configured to determine the purchase probabilities of a customer purchasing multiple products to obtain multiple purchase probabilities;

[0062] The second determination unit 402 is configured to take the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type;

[0063] The third determination unit 403 is configured to determine multiple preference types of a target customer when the purchase feature data corresponding to the first purchase type changes, where the purchase feature data is the feature data corresponding to the product with the most purchase times by the customer;

[0064] The first calculation unit 404 is configured to calculate, through a target algorithm model, the preference feature data corresponding to multiple preference types to obtain a calculation result, where the product corresponding to each preference feature data is purchased by the target customer less times than the product corresponding to the purchase feature data;

[0065] The prediction unit 405 is configured to predict the target preference type of the target customer according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than the purchase probability of the product corresponding to the first purchase type, and higher than the purchase probability of the product corresponding to other types purchased by the target customer.

[0066] In summary, for the prediction device for customer purchase preferences provided in the embodiments of the present application, the first determination unit 401 determines the purchase probabilities of multiple products purchased by a customer to obtain multiple purchase probabilities; the second determination unit 402 uses the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; the third determination unit 403 determines multiple preference types of a target customer when the purchase feature data corresponding to the first purchase type changes, where the purchase feature data is the feature data corresponding to the product with the most purchase times by the customer; the first calculation unit 404 calculates, through a target algorithm model, the preference feature data corresponding to multiple preference types to obtain a calculation result, where the product corresponding to each preference feature data is purchased by the target customer less times than the product corresponding to the purchase feature data; the prediction unit 405 predicts the target preference type of the target customer according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than the purchase probability of the product corresponding to the first purchase type, and higher than the purchase probability of the product corresponding to other types purchased by the target customer, solving the problem of inaccurate prediction of customer purchase preferences in the related art. By calculating, through a target algorithm model, the preference feature data corresponding to multiple preference types and predicting the target preference type of the target customer according to the calculation result, the effect of improving the prediction accuracy of customer purchase preferences is achieved.

[0067] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the first determination unit includes: an acquisition module configured to acquire the purchase feature data of different products; a first calculation module configured to input each purchase feature data into a target algorithm model to calculate the purchase probability corresponding to each product, obtaining multiple purchase probabilities.

[0068] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the first calculation unit includes: a first determination module, configured to determine the feature values in each piece of preference feature data; a second calculation module, configured to calculate a first mean value according to the multiple feature values, where the first mean value is used to represent the average value of the multiple feature values; and a second determination module, configured to use the first mean value as one of the calculation results.

[0069] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the device further includes: a second calculation unit, configured to subtract each feature value from the first mean value to obtain multiple first differences before predicting the target preference type of the target customer according to the calculation results; and a fourth determination unit, configured to use the minimum value among the multiple first differences as the first target difference.

[0070] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the first calculation unit includes: an output module, configured to input multiple pieces of preference feature data into a target algorithm model and output multiple pieces of target preference feature data; a third determination module, configured to determine the target feature values in each piece of target preference feature data; a third calculation module, configured to calculate a second mean value according to the multiple target feature values, where the second mean value is used to represent the average value of the multiple target feature values; and a fourth determination module, configured to use the second mean value as one of the calculation results.

[0071] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the device further includes: a third calculation unit, configured to subtract the feature value in each piece of target preference feature data from the second mean value to obtain multiple second differences before predicting the target preference type of the target customer according to the calculation results; a fifth determination unit, configured to use the minimum value among the multiple second differences as the second target difference; a sixth determination unit, configured to determine a target probability according to the first target difference and the second target difference; and a seventh determination unit, configured to determine the target preference type according to the target probability.

[0072] Optionally, in the prediction device for customer purchase preferences provided in the embodiments of the present application, the sixth determination unit includes: a sorting module, configured to perform priority sorting on multiple preference types through the first target difference and the second target difference to obtain the sorted preference types; a fifth determination module, configured to determine a target matrix according to the sorted preference types; and a fourth calculation module, configured to calculate a target probability by calculating the target matrix.

[0073] The prediction device for customer purchase preferences includes a processor and a memory. The above-mentioned first determination unit 401, second determination unit 402, third determination unit 403, first calculation unit 404, prediction unit 405, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to implement corresponding functions.

[0074] The processor contains a kernel, which retrieves the corresponding program unit from the memory. One or more kernels can be set, and the customer purchase preference is predicted by adjusting the kernel parameters.

[0075] The memory may include non-permanent memory in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. The memory includes at least one memory chip.

[0076] An embodiment of the present invention provides a computer-readable storage medium with a program stored thereon. When the program is executed by a processor, it implements a method for predicting customer purchase preferences.

[0077] An embodiment of the present invention provides a processor for running a program. When the program runs, it executes a method for predicting customer purchase preferences.

[0078] As Figure 5 shown, an embodiment of the present invention provides an electronic device. The device includes a processor, a memory, and a program stored on the memory and executable on the processor. When the processor executes the program, the following steps are implemented: determining the purchase probabilities of a customer purchasing multiple products to obtain multiple purchase probabilities; taking the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; when the purchase characteristic data corresponding to the first purchase type changes, determining multiple preference types of the target customer, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchases by the customer; calculating a calculation result through a target algorithm model for the preference characteristic data corresponding to the multiple preference types, where the product corresponding to each preference characteristic data is purchased by the target customer less times than the product corresponding to the purchase characteristic data; predicting the target preference type of the target customer according to the calculation result, where the product corresponding to the target preference type is purchased by the target customer with a purchase probability lower than that of the product corresponding to the first purchase type and higher than the purchase probability of the product corresponding to other types purchased by the target customer.

[0079] When the processor executes the program, the following steps are also implemented: collecting the purchase characteristic data of different products; inputting each purchase characteristic data into the target algorithm model to calculate the purchase probability corresponding to each product to obtain multiple purchase probabilities.

[0080] When the processor executes the program, the following steps are also implemented: determining the characteristic values in each preference characteristic data; calculating a first mean value according to the multiple characteristic values, where the first mean value is used to represent the average value of the multiple characteristic values; taking the first mean value as one of the calculation results.

[0081] When the processor executes the program, the following steps are also implemented: before predicting the target preference type of the target customer according to the calculation result, subtract each feature value from the first mean value to obtain a plurality of first differences; use the minimum value among the plurality of first differences as the first target difference.

[0082] When the processor executes the program, the following steps are also implemented: input a plurality of preference feature data into the target algorithm model to output a plurality of target preference feature data; determine the target feature value in each target preference feature data; calculate a second mean value according to the plurality of target feature values, where the second mean value is used to represent the average value of the plurality of target feature values; use the second mean value as one of the calculation results.

[0083] When the processor executes the program, the following steps are also implemented: before predicting the target preference type of the target customer according to the calculation result, subtract the feature value in each target preference feature data from the second mean value to obtain a plurality of second differences; use the minimum value among the plurality of second differences as the second target difference; determine the target probability according to the first target difference and the second target difference; determine the target preference type according to the target probability.

[0084] When the processor executes the program, the following steps are also implemented: prioritize a plurality of preference types through the first target difference and the second target difference to obtain the sorted preference types; determine the target matrix according to the sorted preference types; calculate the target probability by calculating the target matrix.

[0085] The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0086] This application also provides a computer program product, which, when executed on a data processing device, is adapted to execute a program initialized with the following method steps: determine the purchase probabilities of a customer purchasing multiple products to obtain a plurality of purchase probabilities; use the type of the product with the highest purchase probability among the plurality of purchase probabilities as the first purchase type; when the purchase feature data corresponding to the first purchase type changes, determine a plurality of preference types of the target customer, where the purchase feature data is the feature data corresponding to the product with the most purchase times by the customer; calculate a calculation result through the target algorithm model for the preference feature data corresponding to the plurality of preference types, where the product corresponding to each preference feature data is purchased by the target customer less times than the product corresponding to the purchase feature data; predict the target preference type of the target customer according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than that of the product corresponding to the first purchase type and higher than that of the products corresponding to other types purchased by the target customer.

[0087] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: collecting purchase characteristic data of different products; inputting each purchase characteristic data into a target algorithm model to calculate the purchase probability corresponding to each product, and obtaining multiple purchase probabilities.

[0088] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: determining the characteristic values in each preference characteristic data; calculating a first mean value according to multiple characteristic values, where the first mean value is used to represent the average value of multiple characteristic values; taking the first mean value as one of the calculation results.

[0089] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: before predicting the target preference type of a target customer according to the calculation results, subtracting each characteristic value from the first mean value to obtain multiple first differences; taking the minimum value among the multiple first differences as the first target difference.

[0090] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: inputting multiple preference characteristic data into a target algorithm model to output multiple target preference characteristic data; determining the target characteristic values in each target preference characteristic data; calculating a second mean value according to multiple target characteristic values, where the second mean value is used to represent the average value of multiple target characteristic values; taking the second mean value as one of the calculation results.

[0091] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: before predicting the target preference type of a target customer according to the calculation results, subtracting the characteristic value in each target preference characteristic data from the second mean value to obtain multiple second differences; taking the minimum value among the multiple second differences as the second target difference; determining the target probability according to the first target difference and the second target difference; determining the target preference type according to the target probability.

[0092] When executed on a data processing device, it is also suitable for executing a program initialized with the following method steps: prioritizing multiple preference types through the first target difference and the second target difference to obtain the sorted preference types; determining a target matrix according to the sorted preference types; calculating the target probability by calculating the target matrix.

[0093] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0094] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0095] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0096] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0097] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0098] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0099] A computer-readable medium includes both permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information accessible by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0100] It should also be noted that the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, such that a process, method, article or apparatus comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or apparatus. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or apparatus comprising the element.

[0101] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, system or computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can 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.

[0102] The above are only embodiments of the present application and are not used to limit the present application. For those skilled in the art, the present application can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A method for predicting customer purchase preferences, characterized in that, it includes: Determine the purchase probabilities of a customer purchasing multiple products to obtain multiple purchase probabilities; Take the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; When the purchase characteristic data corresponding to the first purchase type changes, determine multiple preference types of the target customer, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchases by the customer; Calculate the preference characteristic data corresponding to the multiple preference types through a target algorithm model to obtain a calculation result, where the product corresponding to each preference characteristic data is purchased by the target customer fewer times than the product corresponding to the purchase characteristic data; Calculating the preference characteristic data corresponding to the multiple preference types through a target algorithm model to obtain a calculation result includes: determining the characteristic values in each preference characteristic data; calculating a first mean value according to the multiple characteristic values, where the first mean value is used to represent the average value of the multiple characteristic values; taking the first mean value as one of the calculation results; Subtract each characteristic value from the first mean value to obtain multiple first differences; take the minimum value among the multiple first differences as the first target difference; Calculating the preference characteristic data corresponding to the multiple preference types through a target algorithm model to obtain a calculation result further includes: inputting multiple preference characteristic data into the target algorithm model and outputting multiple target preference characteristic data; determining the target characteristic values in each target preference characteristic data; calculating a second mean value according to the multiple target characteristic values, where the second mean value is used to represent the average value of the multiple target characteristic values; Taking the second mean value as one of the calculation results; subtracting the characteristic value in each target preference characteristic data from the second mean value to obtain multiple second differences; taking the minimum value among the multiple second differences as the second target difference; Predict the target preference type of the target customer according to the calculation result, including: determining a target probability according to the first target difference and the second target difference; determining the target preference type according to the target probability, where the product corresponding to the target preference type is purchased by the target customer with a purchase probability lower than that of the product corresponding to the first purchase type and higher than the purchase probability of the products corresponding to other types purchased by the target customer.

2. The method according to claim 1, characterized in that, Determining the purchase probabilities of a target customer purchasing multiple products to obtain multiple purchase probabilities includes: Collect the purchase characteristic data of different products; Input each purchase characteristic data into the target algorithm model to calculate the purchase probability corresponding to each product to obtain multiple purchase probabilities.

3. The method according to claim 1, characterized in that, Determining the target probability according to the first target difference and the second target difference includes: Rank the multiple preference types according to the first target difference and the second target difference to obtain the ranked preference types; Determine a target matrix according to the ranked preference types; The target probability is obtained by calculating the target matrix.

4. A prediction device for customer purchase preferences, characterized in that, it includes: A first determination unit, configured to determine the purchase probabilities of a customer purchasing multiple products, and obtain multiple purchase probabilities; A second determination unit, configured to use the type of the product with the highest purchase probability among the multiple purchase probabilities as the first purchase type; A third determination unit, configured to determine multiple preference types of a target customer in the case where the purchase characteristic data corresponding to the first purchase type changes, where the purchase characteristic data is the characteristic data corresponding to the product with the most purchase times by the customer; A first calculation unit, configured to calculate, through a target algorithm model, the preference characteristic data corresponding to the multiple preference types to obtain a calculation result, where the product corresponding to each preference characteristic data is purchased by the target customer less times than the product corresponding to the purchase characteristic data; The first calculation unit includes: a first determination module, configured to determine the characteristic values in each preference characteristic data; A second calculation module, configured to calculate a first mean value according to multiple characteristic values, where the first mean value is used to represent the average value of the multiple characteristic values; a second determination module, configured to use the first mean value as one of the calculation results; The device further includes: a second calculation unit, configured to subtract each characteristic value from the first mean value to obtain multiple first differences before predicting the target preference type of the target customer according to the calculation result; a fourth determination unit, configured to use the minimum value among the multiple first differences as the first target difference; The first calculation unit includes: an output module, configured to input the multiple preference characteristic data into the target algorithm model and output multiple target preference characteristic data; a third determination module, configured to determine the target characteristic values in each target preference characteristic data; a third calculation module, configured to calculate a second mean value according to the multiple target characteristic values, where the second mean value is used to represent the average value of the multiple target characteristic values; a fourth determination module, configured to use the second mean value as one of the calculation results; The device further includes: a third calculation unit, configured to subtract the characteristic value in each target preference characteristic data from the second mean value to obtain multiple second differences before predicting the target preference type of the target customer according to the calculation result; a fifth determination unit, configured to use the minimum value among the multiple second differences as the second target difference; A prediction unit, configured to predict the target preference type of the target customer according to the calculation result, where the purchase probability of the product corresponding to the target preference type being purchased by the target customer is lower than the purchase probability of the product corresponding to the first purchase type, and higher than the purchase probability of the product corresponding to other types purchased by the target customer; A sixth determination unit, configured to determine a target probability according to the first target difference and the second target difference; a seventh determination unit, configured to determine the target preference type according to the target probability.

5. A computer-readable storage medium, characterized in that, the storage medium stores a program, where the program executes the method according to any one of claims 1 to 3.

6. An electronic device, characterized in that, Comprising one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the method according to any one of claims 1 to 3.

Images