Detailed Description
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in this application and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It is to be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present application. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.
In the embodiment of the application, under the condition that the historical data of the candidate object is limited, the target user of each candidate object is obtained, the proportion information of the target user of each candidate object in all the users in the corresponding dimension is obtained from different dimensions, then the fraction of the corresponding candidate object is obtained according to the proportion information, and finally the target object is determined by combining a traditional prediction model, so that the accuracy of prediction is improved. The following describes in detail an implementation process of the present application with reference to specific embodiments.
Fig. 1 is a flowchart illustrating a method for determining a target object according to an exemplary embodiment of the present application, which is described from a third party payment platform side. As shown in fig. 1, the method for determining a target object includes:
step S101, obtaining the proportion information of the target user of each candidate object in all the users in the corresponding dimension from different dimensions, and obtaining the score of each candidate object according to the proportion information.
The candidate objects may be merchants which are just established soon, and the target user of each candidate object may be a user whose business frequency reaches a preset frequency within an initial preset time established for the corresponding candidate object, where the preset time and the preset frequency may be set as needed, for example, the preset time may be half a year, and the preset frequency may be 2 times, 3 times, and the like, that is, the target user may be a user with foresight, and they usually start shopping at the merchant and may find a merchant with a better business mode from the user at the initial establishment stage of the merchant.
The different dimensions may include, but are not limited to, a business times dimension, a profession dimension, a user number dimension, and the like. In this embodiment, obtaining the proportion information of the target user of each candidate object in all the users in the corresponding dimension from different dimensions may include: acquiring the proportion of target users with excessive service records in all target users corresponding to the candidate objects from the service frequency dimension; acquiring the proportion of the target user in all users corresponding to the candidate object from the quantity dimension; and acquiring the proportion of the target user corresponding to the candidate object in all users in the industry of the city from the occupation dimension.
And obtaining the score of each candidate object according to the ratio information may include: and obtaining the score of the corresponding candidate object according to the proportion of the target user with excessive service records in all the target users of the corresponding candidate object, the proportion of the target user in all the users of the corresponding candidate object and the maximum value of the proportion of the target user of the corresponding candidate object in all the users of the industry of the city.
For convenience of description, the proportion of the target users in all users corresponding to the candidate object may be represented by Y1, the proportion of the target users having too many shopping records in the candidate object in all target users of the candidate object may be represented by Y2, and the proportion of the target users in the candidate object in all users in the industry belonging to the city may be represented by Y3, and then the score of the corresponding candidate object may be obtained by using a formula f (Y1Y 2 Max (Y3)), where Y represents the identifier of the candidate object.
In this embodiment, the maximum ratio of the target users to all users in the industry of the city is obtained because the merchants with the highest ratio of the target users to all users in the industry of the city have wide user acceptance in the profession of the city, and thus, the target users have a high opportunity to be copied to users in the same profession in other cities.
Suppose that the target user of candidate 1 is users 1-3, all users of candidate 1 are users 1-100, user 1 has two shopping records, users 2-3 have one shopping record, the cities to which users 1-3 belong are Shanghai, the industry to which user 1 belongs is lawyer, the industry to which user 2 belongs is teacher, the industry to which user 3 belongs is lawyer, and lawyers in shanghai are 400 and teachers are 500, the target user accounts for 3% of all users of candidate 1, the proportion of the target users having many shopping records in the candidate object 1 among all the target users is 1/3, the maximum value of the proportion of the target users in the candidate object 1 among all the users in the industry of the city is 1/200, and after the data are obtained, the score of the candidate object 1 can be calculated.
Step S102, acquiring the user amount of each candidate object in a future preset time period.
In this embodiment, step S102 may include: inputting the variable information of each candidate object into a prediction model to obtain the growth speed of each candidate object in a future preset time period; and then, according to the current user quantity of each candidate object and the corresponding growth speed thereof, calculating the user quantity of each candidate object in a future preset time period.
The variable information of the candidate object may include, but is not limited to, one or more of the following information:
the method comprises the following steps of current user scale, transaction number, transaction amount, average value of increasing ring ratios in the last 6 months, user retention rate, new user ratio, each occupation permeability, average value of increasing of occupation permeability ranked at the top 3%, each city ratio, city permeability, average value of increasing of city permeability ranked at the top 3%, county-city ratio, county-city permeability and average value of increasing of county-city permeability.
The predictive model may include, but is not limited to, an iterative Decision Tree (GBDT) model.
For example, all the variables mentioned above may be input into the GBDT model, and the probability that the candidate object, for example, the candidate merchant increases speed by more than 70% in the next 3 months is obtained, and assuming that the probability that the candidate object 1 increases speed by more than 70% in the next 3 months is 10%, the user amount of the candidate object 1 in the next 3 months is: current user volume 1.07.
Therefore, the user quantity of each candidate object in the future preset time period can be obtained through the GBDT model.
Step S103, determining the candidate objects with the scores reaching the preset threshold value and the user amount reaching the preset number as target objects.
In this embodiment, the score ranking and the user amount have respective preset thresholds, for example, a candidate object with a score ranking of the top 5% and a user amount of 500 people may be determined as the target object.
According to the embodiment, the proportion information of the target user of each candidate object in all the users in the corresponding dimension is obtained from different dimensions, the score of each candidate object is obtained according to the proportion information, and the target object is determined by combining the prediction model, so that the defect that the accuracy of determining the target object is low only by the prediction model under the condition that the historical data of the candidate object is limited is overcome, and the prediction accuracy is greatly improved.
Fig. 2A is a flowchart illustrating a process of obtaining a target user of each candidate object according to an exemplary embodiment of the present application, where as shown in fig. 2A, the process of obtaining a target user of each candidate object includes:
step S201, counting the user service information of each candidate object according to the order data.
The user service information of each candidate object may include the first shopping time and the frequency of purchases of the corresponding user at the current candidate object.
Step S202, potential scores of each candidate object are obtained.
The potential score is used to represent a predicted value of the development potential of the corresponding candidate object, and the potential score can be set for each candidate object by experts according to experience.
And step S203, determining a target user for each candidate object according to the user service information and the potential score.
As shown in fig. 2B, step S203 may include the following steps:
step S2031, for each candidate object, calculating an index of each user with respect to the current candidate object according to the user service information and weight thereof of the current candidate object, the potential score and weight thereof of the current candidate object, and the score assigned to the current candidate object by each user.
For convenience of description, in this embodiment, the first time shopping time of each user for the current candidate object may be represented by X1, the potential score of the current candidate object may be represented by X2, the purchase frequency of each user for the current candidate object may be represented by X3, the score assigned to the current candidate object by each user may be represented by i, and the weights of X1, X2 and X3 may be represented by a, b and c, respectively, where the sizes of a, b and c may also be set by experts, and may be represented by the formula: f (x) ═ Σ (aX1i × bX2i × cX3i), an index of each user with respect to the current candidate object is obtained, where x denotes the identity of the user.
Step S2032, an abnormal user of the current candidate is determined.
In order to eliminate the interference of abnormal users such as cattle, the abnormal users of the current candidate object can be determined in at least one of the following manners in the embodiment:
the first mode is as follows: and determining the users with the ratio of the total business times of all the candidate objects to the total sales times of all the candidate objects reaching a first preset value as abnormal users.
Assuming that there are 100 candidates currently, and the ratio of the total purchase frequency of the 100 candidates to the total sale frequency of the 100 candidates of the user 1 reaches a first preset value, for example, the first 1%, the user 1 is an abnormal user.
The second mode is as follows: and determining the user with the ratio of the total business times of the current candidate object to the total sales times of the current candidate object reaching a second preset value as the abnormal user.
Assuming that the ratio of the total purchase frequency of the current candidate object to the total sale frequency of the current candidate object reaches a second preset value, for example, the first 1%, the user 2 is an abnormal user.
It should be noted that the first preset value and the second preset value may be the same or different.
Step S2033, removing abnormal users from all users of the current candidate object, and determining the users whose indexes of all remaining users reach a preset index threshold as target users.
In order to improve the accuracy of the obtained target users, abnormal users may be removed from all users of the current candidate object, and then, according to the indexes of all remaining users, the users whose index ranking reaches a preset index threshold, for example, the top 5%, are determined as the target users.
The target user of each candidate object can be obtained in the above manner.
It should be noted that, the target user of the candidate object may also change with the change of the user in the candidate object service information, but the target user of the candidate object may be determined through the above steps S201 to S203 no matter how the service information changes. Similarly, a change in the candidate target user may also cause a change in the target object, but the target object may be determined through the above-described steps S101 to S103 regardless of the change in the candidate target user.
According to the embodiment, the index of each user relative to the current candidate object is comprehensively calculated according to a plurality of factors such as the service information and the weight of each user in the current candidate object, the potential score and the weight of the current candidate object and the score distributed by each user to the current candidate object, and then after abnormal users are removed from all users of the current candidate object, the target user is determined according to the index ranking of all the remaining users, so that the high accuracy is achieved, and conditions are provided for subsequently determining the target object.
Fig. 3 is a schematic diagram of another process for determining a target object according to an exemplary embodiment of the present application, where as shown in fig. 3, the process for determining a target object may include:
s301, inputting the variable information of each candidate object into a prediction model to obtain the growth speed of each candidate object in a future preset time period.
S302, calculating the user quantity of each candidate object in a future preset time period according to the current user quantity of each candidate object and the corresponding growth speed of each candidate object.
Wherein the future preset time period may be 6 months in the future.
S303, determining the target user of each candidate object.
S304, acquiring the proportion information of the target user of each candidate object in all the users in the corresponding dimension from different dimensions, and acquiring the score of each candidate object according to the proportion information.
S305, determining the candidate objects with the scores reaching the preset threshold value and the user amount reaching the preset number as target objects.
For example, a candidate object, whose score reaches the top 6% and the user amount reaches 700 people, may be determined as the target object, i.e., the target object.
Through the embodiment, the process of obtaining the scores of the corresponding candidate objects according to the proportion information of the target user and determining the target object by combining the prediction model can be clearly seen, the process overcomes the defect of low accuracy of determining the target object only by the prediction model under the condition that the historical data of the candidate object is limited, and the prediction accuracy is greatly improved.
Corresponding to the foregoing embodiments of the method for determining a target object, the present application also provides embodiments of an apparatus for determining a target object.
The embodiment of the device for determining the target object can be applied to a third-party payment platform. The device embodiments may be implemented by software, or by hardware, or by a combination of hardware and software. The software implementation is taken as an example, and is formed by reading corresponding computer program instructions in the nonvolatile memory into the memory for operation through the processor of the device where the software implementation is located as a logical means. From a hardware level, as shown in fig. 4, a hardware structure diagram of a device in which an apparatus for determining a target object is located in the present application is shown, except for the processor, the memory, the network interface, and the nonvolatile memory shown in fig. 4, the device in which the apparatus is located in the embodiment may further include other hardware, for example, for a terminal, a camera, a touch screen, a communication component, and the like, and for a server, a forwarding chip responsible for processing a message, and the like, according to an actual function of the device.
Fig. 5 is a block diagram illustrating an apparatus for determining a target object according to an exemplary embodiment of the present application, where the apparatus is applicable to a third party payment platform, and as shown in fig. 5, the apparatus for determining a target object includes: a score obtaining module 51, a user amount obtaining module 52 and a target object determining module 53.
The score obtaining module 51 is configured to obtain, from different dimensions, proportion information of target users of each candidate object in all users of the corresponding dimension, and obtain a score of each candidate object according to the proportion information, where a target user of each candidate object refers to a user whose number of times of service reaches a preset number of times within an initial preset time established for the corresponding candidate object.
The user quantity obtaining module 52 is configured to obtain a user quantity of each candidate object in a future preset time period.
The target object determination module 53 is configured to determine, as the target object, a candidate object in which the score obtained by the score obtaining module 51 reaches a preset threshold and the user amount obtained by the user amount obtaining module 52 reaches a preset number.
In an alternative implementation, as shown in fig. 6A, the apparatus may further include: a statistics module 54, an acquisition module 55, and a determination module 56.
The counting module 54 is configured to count the user service information of each candidate object according to the order data before the score obtaining module obtains the proportion information of the target user of each candidate object in all the users of the corresponding dimension from different dimensions.
The obtaining module 55 is configured to obtain a potential score of each candidate object, where the potential score is used to represent a pre-estimated development potential of the corresponding candidate object.
The determining module 56 is configured to determine a target user for each candidate object according to the user service information counted by the counting module 54 and the potential score obtained by the obtaining module 55.
In an alternative implementation, as shown in fig. 6B, the determining module 56 may include: a calculation submodule 561, a determination submodule 562, and a removal determination submodule 563.
The calculating submodule 561 is configured to calculate, for each candidate object, an index of each user with respect to the current candidate object according to the user service information and the weight thereof of the current candidate object, the potential score and the weight thereof of the current candidate object, and the score assigned to the current candidate object by each user.
The determination sub-module 562 is used to determine the abnormal users of the current candidate.
The removal determining sub-module 563 is configured to remove the abnormal user determined by the determining sub-module 562 from all users of the current candidate object, and determine, as the target user, a user whose index of all remaining users calculated by the calculating sub-module 561 reaches a preset index threshold.
In another alternative implementation, as shown in fig. 6C, the determination sub-module 562 may include: at least one of the first determination unit 5621 and the second determination unit 5622.
The first determining unit 5621 is configured to determine a user who has reached a first preset value in a ratio of the total number of businesses of all the candidate objects to the total number of sales of all the candidate objects as an abnormal user.
The second determining unit 5622 is configured to determine a user whose ratio of the total number of businesses of the current candidate object to the total number of sales of the current candidate object reaches a second preset value as an abnormal user.
In another alternative implementation, as shown in fig. 7, the score obtaining module 51 may include: a first fetch submodule 511, a second fetch submodule 512 and a third fetch submodule 513.
The first obtaining sub-module 511 is configured to obtain, from the service times dimension, a proportion of target users having too many service records in all target users corresponding to the candidate object.
The second obtaining sub-module 512 is configured to obtain, from the number dimension, a proportion of the target user in all users corresponding to the candidate object.
The third obtaining sub-module 513 is configured to obtain, from the occupation dimension, a proportion of all users in an industry to which the city belongs, of the target user corresponding to the candidate object.
In another alternative implementation, as shown in fig. 7, the score obtaining module 51 may further include: a score acquisition sub-module 514.
The score obtaining sub-module 514 is configured to obtain a score of the corresponding candidate object according to a maximum value of a proportion of the target users having the excessive service records, obtained by the first obtaining sub-module 511, in all the target users of the corresponding candidate object, obtained by the second obtaining sub-module 512, and a proportion of the target users of the corresponding candidate object, obtained by the third obtaining sub-module 513, in all the users of the industry to which the city belongs.
In another alternative implementation, as shown in fig. 8, the user quantity obtaining module 52 may include: an input acquisition submodule 521 and a calculation submodule 522.
The input obtaining sub-module 521 is configured to input the variable information of each candidate object into the prediction model, so as to obtain a growth rate of each candidate object in a future preset time period.
The calculating submodule 522 is configured to calculate the user amount of each candidate object in the future preset time period according to the current user amount of each candidate object and the corresponding increase speed obtained by the input obtaining submodule 531.
The implementation process of the functions and actions of each unit in the above device is specifically described in the implementation process of the corresponding step in the above method, and is not described herein again.
For the device embodiments, since they substantially correspond to the method embodiments, reference may be made to the partial description of the method embodiments for relevant points. The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the application. One of ordinary skill in the art can understand and implement it without inventive effort.
According to the device for determining the target object, the proportion information of the target user of each candidate object in all the users in the corresponding dimension is obtained from different dimensions, the score of each candidate object is obtained according to the proportion information, and then the target object is determined by combining the prediction model, so that the defect that the accuracy rate of determining the target object is low only by depending on the prediction model under the condition that the historical data of the candidate object is limited is overcome, and the prediction accuracy is greatly improved.
Other embodiments of the present application will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the application and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the application being indicated by the following claims.
It will be understood that the present application is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.