CN108961071B

CN108961071B - Method for automatically predicting combined service income and terminal equipment

Info

Publication number: CN108961071B
Application number: CN201810554475.3A
Authority: CN
Inventors: 汪欢
Original assignee: Ping An Life Insurance Company of China Ltd
Current assignee: Ping An Life Insurance Company of China Ltd
Priority date: 2018-06-01
Filing date: 2018-06-01
Publication date: 2023-07-21
Anticipated expiration: 2038-06-01
Also published as: CN108961071A

Abstract

The invention is applicable to the technical field of data processing, and provides a method for automatically predicting combined service benefit, terminal equipment and a computer readable storage medium, wherein the method comprises the following steps: obtaining a plurality of groups of learning samples, and processing the plurality of groups of learning samples through a preset processing algorithm to obtain a service selection model, wherein each group of learning samples consists of a main service sample, an additional service sample and a user sample; inputting a main service type and a characteristic value of a user object as prediction object information into the service selection model, and combining an additional service type output by the service selection model and the main service type into a target combined service of the user object; and determining a target prediction strategy corresponding to the target combined service from a plurality of sets of prediction strategies associated with the main service type, and acquiring a benefit value of the target combined service based on the target prediction strategy. The invention improves the reliability of predicting the target combined service and predicting the profit value.

Description

Method for automatically predicting combined service income and terminal equipment

Technical Field

The invention belongs to the technical field of data processing, and particularly relates to a method for automatically predicting combined service benefits, terminal equipment and a computer readable storage medium.

Background

Business refers to a related transaction that is sold, while benefits are those that can be achieved by business activity. Before a service is developed, the profit of the service is usually predicted, and the prediction method is to acquire relevant characteristics of the service and calculate according to the calculation logic fixed by the service. By way of an insurance example, when an applicant performs insurance on a dangerous species, the benefits obtainable by the insurance are calculated according to the relevant information of the applicant.

As business activities tend to be increasingly complex, a composite business may include a main business and multiple additional businesses, where the combination of the main business and an additional business may need to be calculated in calculating revenue, such as where an insurance includes a main insurance and an additional insurance. Under the condition that the user selects the main service and does not select the additional service, if the traditional method is adopted for calculation, the benefit value cannot be calculated due to the lack of the additional service, only the combination of each additional service and the main service can be calculated independently, much calculated data is not needed, and the reliability of benefit prediction is low.

Disclosure of Invention

In view of this, embodiments of the present invention provide a method, a terminal device, and a computer readable storage medium for automatically predicting benefit of a combined service, so as to solve the problem in the prior art that reliability of benefit prediction for a combined service is low under the condition of known main service.

A first aspect of an embodiment of the present invention provides a method for automatically predicting a combined service benefit, including:

obtaining a plurality of groups of learning samples, and processing the plurality of groups of learning samples through a preset processing algorithm to obtain a service selection model, wherein each group of learning samples consists of a main service sample, an additional service sample and a user sample;

inputting a main service type and a characteristic value of a user object as prediction object information into the service selection model, and combining an additional service type output by the service selection model and the main service type into a target combined service of the user object;

and determining a target prediction strategy corresponding to the target combined service from a plurality of sets of prediction strategies associated with the main service type, and acquiring a benefit value of the target combined service based on the target prediction strategy.

A second aspect of an embodiment of the present invention provides a terminal device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the following steps when executing the computer program:

A third aspect of the embodiments of the present invention provides a computer readable storage medium storing a computer program which when executed by a processor performs the steps of:

Compared with the prior art, the embodiment of the invention has the beneficial effects that:

according to the method, the device and the system, a plurality of groups of learning samples are obtained, each group of learning samples consists of a main service sample, an additional service sample and a user sample, the plurality of groups of learning samples are calculated through a preset processing algorithm, a service selection model is built according to a calculation result, characteristic values of the main service type and the user object are input into the service selection model as prediction object information, the additional service type and the main service type output by the service selection model are combined into a target combined service of the user object, finally, a target prediction strategy corresponding to the target combined service is determined from a plurality of preset prediction strategies related to the main service type, and a profit value of the target combined service is calculated and obtained based on the target prediction strategy.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of an implementation of a method for automatically predicting combined service benefit according to an embodiment of the present invention;

FIG. 2 is a flowchart of a method for automatically predicting combined service benefit according to a second embodiment of the present invention;

FIG. 3 is a flowchart of a method for automatically predicting combined service benefit according to a third embodiment of the present invention;

FIG. 4 is a flowchart of a method for automatically predicting the benefit of a combined service according to the fourth embodiment of the present invention;

FIG. 5 is a flowchart of a method for automatically predicting combined service benefit according to a fifth embodiment of the present invention;

fig. 6 is a block diagram of a terminal device according to a sixth embodiment of the present invention;

fig. 7 is a schematic diagram of a terminal device according to a seventh embodiment of the present invention;

FIG. 8 is a schematic diagram of a service selection model according to an eighth embodiment of the present invention;

Fig. 9 is a schematic diagram of another service selection model provided in a ninth embodiment of the present invention.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, techniques, etc., in order to provide a thorough understanding of the embodiments of the present invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.

In order to illustrate the technical scheme of the invention, the following description is made by specific examples.

Fig. 1 shows an implementation flow of a method for automatically predicting a combined service benefit according to an embodiment of the present invention, which is described in detail below:

in S101, a plurality of groups of learning samples are obtained, and the plurality of groups of learning samples are processed by a preset processing algorithm to obtain a service selection model, where each group of learning samples is composed of a main service sample, an additional service sample and a user sample.

The service refers to a transaction work to be processed, and in the embodiment of the present invention, the service is a combined service, and is composed of a main service and an additional service, where the main service type and the additional service type indicate the types of the main service and the additional service, respectively. In the traditional method, on the premise of knowing the main service which the user needs to transact, more additional services which the user can transact on the basis of the main service are obtained through subjective judgment. In the embodiment of the present invention, a plurality of groups of learning samples are first obtained, one group of learning samples is related information of the ongoing or completed combined service, each group of learning samples is composed of a main service sample, an additional service sample and a user sample, wherein it is worth mentioning that the main service type and the additional service type are names facing the user object in the embodiment, and the main service sample and the additional service sample in each group of learning samples have substantially the same meaning as the main service type and the additional service type respectively, and the names are different only for convenience in distinguishing the learning samples from the user objects.

The user sample in each group of learning samples is a description of a user handling the combined service corresponding to the learning sample, the user sample comprises a plurality of feature values corresponding to a plurality of known features one by one, the plurality of known features can be preset according to specific services, for example, the plurality of known features are gender, age and wages, and the feature values indicate the attributes of the corresponding known features. Optionally, for facilitating calculation, before processing the multiple groups of learning samples through a preset processing algorithm, performing interval reassignment operation on multiple feature values of the user samples in each group of learning samples, and assigning a corresponding calculated value to each feature value. For example, if a certain known feature is age, the interval may be divided into less than fifty years old and equal to or more than fifty years old, and if the feature value of the known feature in the user sample is less than fifty years old, the calculated value of the known feature in the user sample is assigned 0; if the feature value of the known feature in the user sample is greater than or equal to fifty years old, the calculated value of the known feature in the user sample is assigned to 1, and of course, the division of the interval and the assigned calculated value may be set according to the actual application scenario, that is, the interval reassignment operation is not limited to the above example.

Optionally, multiple sets of learning samples are acquired according to preset acquisition conditions. In the embodiment of the invention, the acquisition conditions can be preset, a plurality of groups of learning samples are screened from a large number of service samples according to the acquisition conditions so as to qualitatively predict the benefit of the combined service, wherein the service samples and the learning samples have the same format, which means all relevant information related to the combined service in the embodiment of the invention, and the number of the service samples is usually more. The specific acquisition process is as follows: if the profit prediction is to be performed on the combined service of the A market, setting the acquisition condition as 'screening the learning sample with the source place being the A market', so as to screen the learning sample from the service samples with a plurality of source places; if a profit prediction is to be performed on a certain known feature, such as a combined business older than fifty years, then the acquisition condition is set to "filter out a learning sample older than fifty years from the user samples". The acquisition condition may be a single condition or a combination of two or more conditions, for example, the acquisition condition may be set to "a learning sample with a source of city a and a user sample age of more than fifty years old is selected from the business samples", and the setting of the acquisition condition is not limited to the above example, and may be determined according to the actual application scenario. In addition, in order to reduce the pressure of acquiring the learning samples and performing subsequent processing, the number of the learning samples may be limited, so that the acquired number of the learning samples is in a preset order of magnitude, and the order of magnitude may be freely set, for example, the order of magnitude may be set to be one thousand.

Optionally, multiple sets of learning samples are obtained from a business database. The service database may be a database for storing the combined service by a main body (such as an organization or a company) running the combined service, generally speaking, the service samples stored in the service database relate to more combined service conditions, and the number of the service samples is high, so that multiple groups of learning samples can be determined directly from the service samples in the service database. In addition, in order to prevent calculation errors caused by deviation in acquisition of learning samples, a random manner is adopted to select a business sample from a business database as the learning sample, or after acquisition conditions are determined, a random manner is adopted to determine the learning sample from business samples which meet the acquisition conditions from the business database.

In the embodiment of the invention, after a plurality of groups of learning samples are obtained, the plurality of groups of learning samples are processed through a preset processing algorithm. The processing modes are divided into two modes, wherein the first mode is to process a main service type sample in each group of learning samples as a known feature, and the main service type sample, an additional service sample and a user sample are taken as input parameters of a service selection model; the second mode is that firstly, multiple groups of learning samples are classified according to the main service samples, the learning samples with the same main service samples are classified into one type, and then each type of learning sample is independently processed, namely, in the second mode, the main service samples are only used as the standard of classifying the multiple groups of learning samples and are not used as input parameters. The process is elucidated on the basis of the second approach as follows:

Firstly, in a plurality of groups of learning samples with the same main business sample, a plurality of diffusion gains of a plurality of known features in a user sample for the plurality of groups of learning samples under the category are calculated respectively, for convenience of explanation, it is assumed that the plurality of known features of the user sample comprise gender Sex, age and payroll, and the Additional business sample comprises Additional _type1 And Additional _type2 Two kinds of them areEach set of learning samples in the class includes a characteristic value of gender set, a characteristic value of Age, a characteristic value of payroll, and one of the two additional business samples. And processing a plurality of groups of learning samples under the class to obtain a service selection model, and firstly, respectively calculating the diffusion gains of the three known features for the plurality of groups of learning samples under the class, wherein the diffusion gains indicate the importance degree of the known features for the plurality of groups of learning samples under the class, and the importance degree is higher as the diffusion gain is higher. In the embodiment of the invention, the diffusion gains of the Sex Sex, the Age and the payroll for all groups of learning samples under the class are calculated at first, and the calculation process can be used for arbitrarily sequencing the calculation sequences of the Sex Sex, the Age and the payroll without limiting the calculation sequences. To illustrate the process of calculating the diffusion gain for gender six.

Obtaining the total Number of the samples of a plurality of groups of learning samples under the category _sample And set Number _sample1 And Number _sample2 The Additional service samples in the groups of learning samples respectively representing the category are Additional _type1 、Additional _type2 If the number of the samples is equal to the predetermined value, a first calculation formula for calculating the overall diffusion gain of the plurality of groups of learning samples under the category is as follows:

in the first calculation formula, K is a preset gain coefficient, so that the obtained overall diffusion gain is convenient for statistics and calculation, for example, K may be set to 10.

Assuming that the calculated values of Sex Sex include two types, sex _type1 And Sex _type2 And assuming Number _type1-sample The calculated value of the known characteristic Sex in the multiple groups of learning samples under the category is Sex _type1 And is at Number _type1-sample In the corresponding learning sample, the Number is assumed _{type1-sample1} For Additional service sample as Additional _type1 Number of samples of (1), assuming Number _{type1-sample2} For Additional service sample as Additional _type2 Is the number of samples; similarly, assume Number _type2-sample The calculated value of the known characteristic Sex in the multiple groups of learning samples under the category is Sex _type2 And is at Number _type2-sample In the corresponding learning sample, the Number is assumed _{type2-sample1} For Additional service sample as Additional _type1 Number of samples of (1), assuming Number _{type2-sample2} For Additional service sample as Additional _type2 Is a number of samples of (a). Then calculate the calculated value Sex of Sex Sex _type1 The second calculation formula of the corresponding diffusion gain is:

calculated value Sex for calculating Sex Sex _type2 The third calculation formula of the corresponding diffusion gain is:

will calculate the value as Sex _type1 And Sex _type2 Regarding as a branch under the gender Sex, a fourth calculation formula for calculating the conditional diffusion gain of the gender Sex with respect to the plurality of groups of learning samples under the category according to the first calculation formula, the second calculation formula and the third calculation formula is:

the conditional diffusion gain obtained by the fourth calculation formula represents the uncertainty of the plurality of groups of learning samples under the condition that the Sex set is determined, so that a fifth calculation formula for calculating the diffusion gain of the Sex set can be obtained according to the overall diffusion gain and the conditional diffusion gain:

Gain _Extended (Sex)＝Gain _Extended (All)-Gain _Extended (All|Sex)

calculated Sex SexDiffusion Gain _Extended (Sex) determines the importance of gender Sex to the sets of learning samples under the category. Similarly, the diffusion Gain of Age can be calculated _Extended (Age) and Salary diffusion Gain _Extended (Salary). Determination of Gain _Extended (Sex)、Gain _Extended (Age) and Gain _Extended (Salary) the greatest diffusion gain and taking the known feature corresponding to the diffusion gain as the first sample node of the service selection model (the first sample node corresponds to all the groups of learning samples under the class), and taking the different calculated values of the known feature as branches to split down. For example, the maximum diffusion Gain is Gain _Extended (Sex), taking two calculated values of the Sex Sex as two branches, splitting a plurality of groups of learning samples under the category downwards, wherein the calculated value corresponding to the characteristic value of the Sex of one of split sample nodes is Sex _type1 Another sample node after splitting is a characteristic value of Sex corresponding to a calculated value of Sex _type2 Is a learning sample of (a). After splitting, continuously calculating a plurality of diffusion gains of a plurality of known features except the gender Sex on the learning sample corresponding to the split sample node for the split sample node, taking the calculated value of the known feature corresponding to the diffusion gain with the largest value as a branch, splitting the split sample node again, and continuously repeating the process until all the known features of the user sample are taken as the sample nodes, thus finishing the construction of the service selection model. FIG. 8 is a schematic view of a service selection model calculated in the second manner, and the Age has a calculated value Age as shown in FIG. 8 _type1 And Age _type2 Salary has a calculated value Salary _type1 And Salary _type2 Sex Sex is the first level sample node in the service selection model, age and payroll are the second level sample node, the lowest level is the Additional service sample obtained by different calculated values, including Additional _type1 And Additional _type2 . In the sample node of Age, the calculated value of Sex Sex is Sex _type1 Multiple sets of learning patternsThe cost is high. It should be noted that if there are N different main service samples, i.e. there are N classified learning samples, then there are N service selection models constructed.

In addition, fig. 9 is a schematic diagram of another service selection model calculated in the first manner, and as shown in fig. 9, the main service sample includes two types of sample one and sample two, and the Sex Sex has a calculated value Sex _type1 And Sex _type2 Age has calculated Age _type1 And Age _type2 Salary has a calculated value Salary _type1 And Salary _type2 The Additional service sample comprises Additional _type1 、Additional _type2 And Additional _type3 Three types. FIG. 9 shows that the Additional service samples of all the groups of learning samples corresponding to the main service sample as sample two are Additional _type3 After the branch structure of the main service sample as the sample two is obtained, the branch structure is not split downwards.

In S102, the main service type and the feature value of the user object are input as prediction object information into the service selection model, and the additional service type output by the service selection model and the main service type are combined into the target combined service of the user object.

After the service selection model is generated, acquiring the main service type and the characteristic value of the user object, and inputting the main service type and the characteristic value of the user object into the service selection model as prediction object information, wherein the characteristic value of the user object accords with the format of a user sample, for example, the characteristic value of gender, the characteristic value of age and the characteristic value of wages are included in the user sample, and the characteristic value of the user object also needs to include the characteristic value of gender, the characteristic value of age and the characteristic value of wages. According to the different generation modes of the service selection model, after the prediction object information is input into the service selection model, the calculation modes are different, if the service selection model is generated by adopting the first mode, namely, a main service sample in a learning sample is used as an input parameter, in the step, the prediction object information is directly calculated through the service selection model to obtain an additional service type; if the service selection model is generated by adopting the second mode, selecting the service selection model corresponding to the main service type in the predicted object information from the plurality of service selection models, and inputting the characteristic value of the user object into the service selection model so as to obtain the additional service type output by the service selection model. And after the additional service type output by the service selection model is obtained, combining the additional service type and the main service type into a target combined service.

In S103, a target prediction policy corresponding to the target combined service is determined from among a plurality of sets of prediction policies associated with the main service type, and a benefit value of the target combined service is obtained based on the target prediction policy.

For one main traffic type it may be combined with different additional traffic types, thus forming different combined traffic, and the manner of revenue calculation for the different combined traffic may also be different. A plurality of prediction strategies of a plurality of combined services associated with the main service type are obtained, wherein the prediction strategies can be an arithmetic expression, a calculation logic or the like. And according to the target combined service determined in the step S102, determining a target prediction strategy corresponding to the target combined service from a plurality of sets of prediction strategies, and calculating the target combined service based on the target prediction strategy to obtain a benefit value. It should be noted that, since the target combined service may only relate to the type of service, and does not include specific service parameters, the service parameters corresponding to the target combined service are acquired before calculating the benefit value. The calculation of the target combined service is actually a process of calculating the service parameters based on the target prediction strategy. Taking a target combined service as a combined insurance service example, wherein the combined insurance service comprises a main insurance type and an auxiliary insurance type, the target prediction strategy corresponding to the combined insurance service takes the premium of the auxiliary insurance as an intermediate premium, and the sum of the intermediate premium and the premium of the main insurance is taken as the premium of the main insurance, and the calculated premium of the main insurance is the benefit value of the combined insurance service. And after the target prediction strategy is determined, acquiring the premium of the additional insurance and the premium of the main insurance as service parameters.

In the embodiment of the invention, the target combined service is unique and has no inclusion relationship, for example, the target combined service with the main service type A and the additional service type B has no inclusion relationship with the target combined service with the main service type A and the additional service types B and C, and the target combined service is two independent combined services.

The method comprises the steps of storing a plurality of sets of prediction strategies in an EXCEL table or a database, setting a script file, automatically acquiring a target prediction strategy corresponding to a target combination service from the EXCEL table or the database after the target combination service is detected to be determined in the process of executing the script file, and calculating the profit value of the target combination service according to the target prediction strategy, so that the automation degree of profit calculation is improved.

As can be seen from the embodiment shown in fig. 1, in the embodiment of the present invention, by acquiring multiple groups of learning samples, each group of learning samples is composed of a main service sample, an additional service sample and a user sample, calculating the multiple groups of learning samples by a processing algorithm to obtain a service selection model, then inputting the characteristic values of the main service type and the user object as prediction object information into the service selection model, combining the additional service type and the main service type output by the service selection model into a target combined service of the user object, and finally determining a target prediction strategy corresponding to the target combined service from multiple sets of prediction strategies associated with the main service type, and acquiring a benefit value of the target combined service based on the target prediction strategy.

Fig. 2 shows an implementation method obtained by refining the steps of processing multiple groups of learning samples by a processing algorithm on the basis of the first embodiment of the present invention. The embodiment of the invention provides an implementation flow chart of a method for automatically predicting the benefit of combined service, as shown in fig. 2, the method for automatically predicting the benefit of combined service can comprise the following steps:

in S201, a preset sample value range corresponding to the user sample is obtained.

In the process of generating and storing the learning samples, human errors or program errors may occur, so that abnormal samples may exist in the acquired multiple groups of learning samples, and because the abnormal samples are generally abnormal in characteristic values of one or a plurality of known characteristics of the user samples, for example, the characteristic values of ages are negative numbers, in the embodiment of the invention, a preset sample value range corresponding to the user samples is acquired, and the sample value range includes a plurality of characteristic value ranges corresponding to a plurality of known characteristics in the user samples. The feature value range may be formulated according to the limitation condition of the known feature by the combined service, for example, if the age of a certain insurance limit applicant is ten years to seventy years, the feature value range corresponding to the age is ten to seventy years.

In S202, learning samples in which the user sample is in the sample value range are selected from the plurality of groups of learning samples, and the selected learning samples are output to a learning sample set.

After the sample value range of the user sample is obtained, detecting a plurality of groups of learning samples based on the sample value range, judging whether the user sample of each group of learning samples is positioned in the sample value range, and specifically judging whether the characteristic values of a plurality of known characteristics of the user sample under each group of learning samples are positioned in the corresponding characteristic value range in the sample value range. And if the user sample of a certain group of learning samples is positioned in the sample value range, outputting the group of learning samples to a learning sample set.

Optionally, if the feature value of a known feature of a user sample in a certain group of learning samples is detected to be empty, an alarm is sent out or the group of learning samples is deleted. If the feature value of a known feature of a user sample in a certain group of learning samples is null, the data loss or the data error may be possible, an alarm prompt may be sent to the user to prompt the user to carry out numerical supplement or numerical correction on the feature value of the known feature of the user sample in the group of learning samples, or to directly delete the group of learning samples. After the user supplements or corrects the numerical value of the user sample in the group of learning samples, the group of learning samples can be uploaded again, and if the feature values of all the known features of the user sample in the group of learning samples are detected to be in the corresponding feature value range, the group of learning samples are automatically output to the learning sample set.

After outputting the learning sample to the learning sample set, repeated samples in the learning sample set may be detected, and as shown in fig. 3, the method for automatically predicting the combined service benefit may include the following steps:

in S301, it is detected whether there are duplicate samples in the learning sample set.

In the actual service, the same learning sample may occur, which means that the main service sample, the additional service sample and the user sample in the learning sample are the same, and the same reason for the occurrence may be that the learning sample is repeatedly stored in advance to prevent data loss; a mere storage error is also possible. Therefore, in the embodiment of the invention, multiple groups of learning samples in the learning sample set are detected, and whether repeated samples exist in the learning sample set is detected, wherein the repeated samples refer to more than two groups of learning samples with the same main service sample, additional service sample and user sample.

In S302, if the repeated samples are detected, only one set of the learning samples among the repeated samples is retained in the learning sample set.

Because the service selection model is calculated according to the number of the learning samples, repeated samples can influence the generation of the subsequent service selection model, so that the fitting degree of the service selection model and the actual situation is reduced. Therefore, in the embodiment of the invention, if the repeated samples are detected, only one group of the repeated samples in the learning sample set is reserved, so that the accuracy of generating the service selection model is effectively improved.

In S203, the learning sample set is processed by the processing algorithm.

After screening a plurality of groups of learning samples, processing the generated learning sample set through a preset processing algorithm to construct a service selection model.

As can be seen from the embodiment shown in fig. 2, in the embodiment of the present invention, a preset sample value range corresponding to a user sample is obtained, a learning sample in which the user sample is in the sample value range is selected from multiple groups of learning samples, the selected learning sample is output to a learning sample set, and finally the learning sample set is processed by a processing algorithm, so that abnormal samples in the multiple groups of learning samples are effectively eliminated, and adverse effects of the abnormal samples on the generation of a service selection model are prevented.

Fig. 4 shows an implementation method of refining a target combination service in which an additional service type and a main service type output by a service selection model are combined into a user object if a plurality of target combination services are required to calculate benefits on the basis of the first embodiment of the present invention. The embodiment of the invention provides an implementation flow chart of a method for automatically predicting the benefit of combined service, as shown in fig. 4, the method for automatically predicting the benefit of combined service can comprise the following steps:

In S401, the additional service type output by the service selection model is added to the additional service set.

In an actual application scenario, after the feature values of the main service type and the user object are determined, a plurality of target combined services need to be obtained, and a plurality of benefit values need to be calculated. Therefore, in the embodiment of the invention, the additional service type output by the service selection model based on the main service type and the characteristic value of the user object is added to the additional service set.

In S402, tracing up in the service selection model based on the additional service type, and acquiring a sample node of the user sample with the highest correlation degree with the additional service type, where the sample node is related to a known feature of the user sample.

After obtaining the additional service type output by the service selection model, since the additional service type is determined by a certain branch structure in the service selection model, tracing up in the service selection model according to the branch structure to which the additional service type belongs, and finding out the sample node of the user sample closest to the additional service type, the sample node corresponding to the known feature in the user sample, the sample node corresponding to the additional service The correlation of the traffic type is also highest. As shown in fig. 9, if the main service type is sample one, the calculated value corresponding to the feature value of the user object is set _type1 And Age _type1 The obtained Additional service type is Additional _type1 And tracing up based on the additional service type, wherein the obtained sample node with the highest correlation degree with the additional service type is Age.

In S403, a secondary service type in the service selection model is acquired, and the secondary service type is added to the additional service set, where the secondary service type is an additional service type subordinate to the sample node and other than the additional service type output by the service selection model.

In the embodiment of the invention, if a plurality of target combined services are required to be generated, after the main service type is determined, the characteristic value with smaller influence degree in the characteristic values of the user object is preferentially considered to be changed, and the additional service type corresponding to the changed characteristic value of the user object in the service selection model is added to the additional service set. Specifically, in the service selection model, each sample node (each known feature of the user sample) may have one or more branch structures under the sub-category, so that the sample node with the highest association with the additional service type in the service selection model is acquired, and other additional service types except for the additional service type output by the service selection model are named as secondary service types for convenience of distinction, and the secondary service types are added to the additional service set. If the sample node with the highest correlation degree is subordinate and does not have other branch structures, namely only the additional service type output by the service selection model exists, or after the secondary service type is added to the additional service set, the preset quantity upper limit of the additional service set is not met, continuing to trace up in the service selection model based on the sample node with the highest correlation degree, determining a sample node with the highest correlation degree at a higher level, and acquiring the secondary service type related to the sample node at the higher level.

In S404, the additional service types and the secondary service types in the additional service set are respectively combined with the main service type into a plurality of the target combined services.

After the additional service set is generated (the generated judgment standard may be that the number of the additional service set reaches the upper limit or other preset conditions are met), the additional service type and the secondary service type in the additional service set are respectively combined with the main service type, so as to obtain a plurality of target combined services.

As can be seen from the embodiment shown in fig. 4, in the embodiment of the present invention, by adding the additional service type output by the service selection model to the additional service set, tracing up in the service selection model based on the additional service type, obtaining a sample node of a user sample with highest correlation with the additional service type, where the sample node is related to a known feature of the user sample, then obtaining a secondary service type subordinate to the sample node in the service selection model except for the additional service type, and finally combining the additional service type and the secondary service type in the additional service set with the main service type into multiple target combined services, where the application scenario that the required target combined service is multiple is applicable, thereby improving the reliability of the multiple target combined services generated under the condition that the multiple combined services need to be subjected to profit prediction.

Fig. 5 shows an implementation method of obtaining a secondary service type in a service selection model and adding the secondary service type to an additional service set for refinement if there is an upper limit on the number of additional service sets and there are multiple secondary service types based on the fourth embodiment of the present invention. The embodiment of the invention provides an implementation flow chart of a method for automatically predicting the benefit of combined service, as shown in fig. 5, the method for automatically predicting the benefit of combined service can comprise the following steps:

in S501, a plurality of gain values of the sample node obtained by the processing algorithm are obtained, where the plurality of gain values are associated with a plurality of the secondary service types one by one.

After determining the sample node with the highest correlation with the additional service types output by the service selection model, the sample node usually has multiple branches in the service selection modelThe structure is characterized in that the additional service types under each branch structure are different, so that a plurality of gain values of the sample node can be calculated based on the branch structure, wherein the gain values are basically the conditional diffusion gains of a plurality of calculated values of the known characteristics corresponding to the sample node relative to a plurality of groups of learning samples respectively. Specifically, the calculated Gain is exemplified by the fourth calculation formula _Extended (ALL|Sex) is the conditional diffusion gain of Sex Sex, then for two classes of calculated values Sex under Sex Sex _type1 And Sex _type2 The corresponding conditional diffusion gains are:

Sex _type1 and Sex _type2 Corresponding two conditional diffusion gains Gain _Extended (ALL|Sex _type1 ) And Gain _Extended (ALL|Sex _type2 ) I.e. two gain values of the known characteristic Sex six. And the same as the calculation method, after determining the sample node (known feature) with the highest correlation degree with the additional service type output by the service selection model, obtaining a plurality of gain values corresponding to a plurality of calculated values related to the known feature one by one.

In S502, a plurality of the secondary service types are ordered according to the values of the gain values, and a secondary service sequence is generated.

Since the conditional diffusion gain represents the influence degree of other conditions on the plurality of groups of learning samples after a certain condition is determined, the larger the conditional diffusion gain is, the smaller the influence degree of the condition corresponding to the conditional diffusion gain on the plurality of groups of learning samples is. The method of sorting the plurality of secondary service types corresponding to the plurality of gain values is preferably from small to large according to the values of the plurality of gain values, wherein each gain value corresponds to a branch structure, that is, corresponds to a secondary service type, in the service selection model. And after the sequencing is completed, generating a secondary service sequence.

In S503, the secondary service types are sequentially selected from the secondary service sequence from front to back, and added to the additional service set until the upper limit of the number of the additional service sets is reached.

After the secondary service sequence is generated, sequentially selecting secondary service types from front to back (namely, from small to large) in the secondary service sequence, adding the secondary service types to the additional service set in sequence, and stopping adding when the number of the additional service set reaches the upper limit of the number. If the number upper limit of the additional service sets is not reached after all the secondary service types in the secondary service sequence are added to the additional service sets, continuing to trace up in the service selection model on the basis of the sample node with the highest correlation degree closest to the additional service types output by the service selection model, obtaining a sample node with a higher level, calculating a plurality of gain values of the sample node with the higher level, generating the secondary service sequence, selecting the secondary service types from front to back in the secondary service sequence, adding the secondary service types to the additional service sets, and iterating the process until the number upper limit of the additional service sets is reached. Optionally, in order to avoid adding duplicate secondary service types, when a secondary service type subordinate to the sample node is acquired or a secondary service type is selected from a secondary service sequence, judging whether the same secondary service type exists in the additional service set, and if so, omitting the secondary service type; if not, the operation of acquiring the secondary service type or selecting the secondary service type is normally executed.

As can be seen from the embodiment shown in fig. 5, in the embodiment of the present invention, by acquiring multiple gain values of a sample node obtained by a processing algorithm, where the multiple gain values are related to multiple secondary service types one by one, and sorting the multiple secondary service types according to the values of the multiple gain values, a secondary service sequence is generated, the secondary service types are sequentially selected from front to back in the secondary service sequence, and the secondary service types are added to an additional service set until the number upper limit of the additional service set is reached, and based on the determination of a sample node, that is, a known feature, the secondary service type most likely to be selected by a user is predicted by calculating the multiple gain values, so that the reliability of generating multiple target combined services is improved.

It should be understood that the sequence number of each step in the foregoing embodiment does not mean that the execution sequence of each process should be determined by the function and the internal logic, and should not limit the implementation process of the embodiment of the present invention.

Fig. 6 shows a block diagram of a terminal device according to an embodiment of the present invention, where the terminal device includes units for performing the steps in the corresponding embodiment of fig. 1. Please refer to fig. 1 and the related description of the embodiment corresponding to fig. 1. For convenience of explanation, only the portions related to the present embodiment are shown.

Referring to fig. 6, the terminal device includes:

the processing unit 61 is configured to obtain multiple groups of learning samples, and process the multiple groups of learning samples through a preset processing algorithm to obtain a service selection model, where each group of learning samples is composed of a main service sample, an additional service sample and a user sample;

a combining unit 62, configured to input a main service type and a feature value of a user object as prediction object information into the service selection model, and combine an additional service type output by the service selection model and the main service type into a target combined service of the user object;

an obtaining unit 63, configured to determine a target prediction policy corresponding to the target combined service from multiple sets of prediction policies associated with the main service type, and obtain a benefit value of the target combined service based on the target prediction policy.

Optionally, the processing unit 61 includes:

the range acquisition unit is used for acquiring a preset sample value range corresponding to the user sample;

the output unit is used for screening out the learning samples of which the user samples are in the sample value range from the plurality of groups of learning samples, and outputting the screened learning samples to a learning sample set;

And the processing subunit is used for processing the learning sample set through the processing algorithm.

Optionally, the output unit further includes:

a detection unit for detecting whether a repeated sample exists in the learning sample set;

and the retaining unit is used for retaining only one group of the learning samples in the repeated samples in the learning sample set if the repeated samples are detected.

Optionally, the combining unit 62 includes:

the first adding unit is used for adding the additional service type output by the service selection model to an additional service set;

the tracing-up unit is used for tracing up in the service selection model based on the additional service type, and obtaining a sample node of the user sample with the highest correlation degree with the additional service type, wherein the sample node is related to the known feature of the user sample;

a second adding unit, configured to obtain a secondary service type in the service selection model, and add the secondary service type to the additional service set, where the secondary service type is an additional service type that belongs to the sample node and is other than the additional service type output by the service selection model;

And the combination subunit is used for respectively combining the additional service types and the secondary service types in the additional service set with the main service type into a plurality of target combined services.

Optionally, if there is an upper limit on the number of additional service sets and there are multiple secondary service types, the second adding unit includes:

a gain obtaining unit, configured to obtain a plurality of gain values of the sample node obtained by the processing algorithm, where the plurality of gain values are related to a plurality of secondary service types one by one;

the sorting unit is used for sorting the plurality of secondary service types according to the numerical values of the plurality of gain values to generate a secondary service sequence;

and the service selection unit is used for sequentially selecting the secondary service types from the secondary service sequence from front to back and adding the secondary service types to the additional service set until the upper limit of the number of the additional service sets is reached.

Therefore, the terminal equipment provided by the embodiment of the invention realizes the prediction of the target combined service and the income value by constructing the service selection model, and improves the reliability of the prediction.

Fig. 7 is a schematic diagram of a terminal device according to an embodiment of the present invention. As shown in fig. 7, the terminal device 7 of this embodiment includes: a processor 70, a memory 71, and a computer program 72 stored in the memory 71 and executable on the processor 70. The processor 70, when executing the computer program 72, implements the steps of the method embodiments for automatically predicting combined service benefit described above, such as steps S101 to S103 shown in fig. 1. Alternatively, the processor 70, when executing the computer program 72, implements the functions of the units in the embodiments of the terminal device described above, for example, the functions of the units 61 to 63 shown in fig. 6.

By way of example, the computer program 72 may be divided into one or more units, which are stored in the memory 71 and executed by the processor 70 to accomplish the present invention. The one or more units may be a series of computer program instruction segments capable of performing a specific function for describing the execution of the computer program 72 in the terminal device 7. For example, the computer program 72 may be divided into a processing unit, a combining unit and an acquisition unit, each unit having the following specific functions:

the processing unit is used for obtaining a plurality of groups of learning samples, processing the plurality of groups of learning samples through a preset processing algorithm to obtain a service selection model, wherein each group of learning samples consists of a main service sample, an additional service sample and a user sample;

the combination unit is used for inputting the main service type and the characteristic value of the user object as prediction object information into the service selection model, and combining the additional service type output by the service selection model and the main service type into a target combined service of the user object;

and the acquisition unit is used for determining a target prediction strategy corresponding to the target combined service from a plurality of sets of prediction strategies associated with the main service type, and acquiring the benefit value of the target combined service based on the target prediction strategy.

The terminal device 7 may be a computing device such as a desktop computer, a notebook computer, a palm computer, a cloud server, etc. The terminal device 7 may include, but is not limited to, a processor 70, a memory 71. It will be appreciated by those skilled in the art that fig. 7 is merely an example of the terminal device 7 and does not constitute a limitation of the terminal device 7, and may include more or less components than illustrated, or may combine certain components, or different components, e.g. the terminal device 7 may further include input-output devices, network access devices, buses, etc.

The processor 70 may be a central processing unit (Central Processing Unit, CPU), or may be another general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), an off-the-shelf programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The memory 71 may be an internal storage unit of the terminal device 7, such as a hard disk or a memory of the terminal device 7. The memory 71 may be an external storage device of the terminal device 7, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card) or the like, which are provided on the terminal device 7. Further, the memory 71 may also include both an internal storage unit and an external storage device of the terminal device 7. The memory 71 is used for storing the computer program as well as other programs and data required by the terminal device 7. The memory 71 may also be used for temporarily storing data that has been output or is to be output.

It will be clearly understood by those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units is illustrated, and in practical application, the above-mentioned functional allocation may be performed by different functional units, that is, the internal structure of the terminal device is divided into different functional units, so as to perform all or part of the above-mentioned functions. The functional units in the embodiment may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit, where the integrated units may be implemented in a form of hardware or a form of a software functional unit. In addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present application. The specific working process of the units in the above system may refer to the corresponding process in the foregoing method embodiment, which is not described herein again.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

In the embodiments provided in the present invention, it should be understood that the disclosed apparatus/terminal device and method may be implemented in other manners. For example, the apparatus/terminal device embodiments described above are merely illustrative, e.g., the division of the units is merely a logical function division, and there may be additional divisions when actually implemented, e.g., multiple units or components may be combined or integrated into another system, or some features may be omitted or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection via interfaces, devices or units, which may be in electrical, mechanical or other forms.

The units described as separate units may or may not be physically separate, and units shown 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 units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

In addition, each functional unit in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the present invention may implement all or part of the flow of the method of the above embodiment, or may be implemented by a computer program to instruct related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the computer program may implement the steps of each of the method embodiments described above. Wherein the computer program comprises computer program code which may be in source code form, object code form, executable file or some intermediate form etc. The computer readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer Memory, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), an electrical carrier signal, a telecommunications signal, a software distribution medium, and so forth. It should be noted that the computer readable medium contains content that can be appropriately scaled according to the requirements of jurisdictions in which such content is subject to legislation and patent practice, such as in certain jurisdictions in which such content is subject to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

The above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention.

Claims

1. A method for automatically predicting combined service revenue, comprising:

Determining a target prediction strategy corresponding to the target combined service from a plurality of sets of prediction strategies associated with the main service type, and acquiring a benefit value of the target combined service based on the target prediction strategy;

the combining the additional service type output by the service selection model and the main service type into the target combined service of the user object comprises the following steps:

adding the additional service type output by the service selection model to an additional service set;

based on the additional service type, tracing upwards in the service selection model, and acquiring a sample node of the user sample with the highest correlation degree with the additional service type, wherein the sample node is related to the known feature of the user sample;

acquiring a secondary service type in the service selection model, and adding the secondary service type to the additional service set, wherein the secondary service type is an additional service type subordinate to the sample node except the additional service type output by the service selection model;

combining the additional service types in the additional service set and the secondary service types with the main service types respectively to form a plurality of target combined services;

Wherein, the business selection model is constructed by the following method:

classifying a plurality of groups of learning samples according to a main service sample, classifying the same learning samples of the main service sample into one type, respectively calculating a plurality of diffusion gains of a plurality of known features of a sample user for the plurality of groups of learning samples under the one type of main service sample in the same plurality of groups of learning samples of the one type of main service sample, determining the largest diffusion gain in the plurality of diffusion gains, taking the known feature corresponding to the largest diffusion gain as a first sample node of a service selection model, taking different calculated values of the known sample features as branches to split downwards, continuing to divide the divided sample nodes into a plurality of diffusion gains of the learning samples corresponding to the divided sample nodes by a plurality of known features except the first sample node after the division, taking the calculated value of the known feature with the largest value as the branch, dividing the divided sample nodes again, and repeating the processes until all the known features of the user sample are taken as sample nodes, and finishing the construction of the service selection model, wherein the diffusion gains indicate that the known features are important to the plurality of groups of samples under the one type of the more important degrees.

2. The method of claim 1, wherein the processing the plurality of sets of learning samples by a preset processing algorithm comprises:

acquiring a preset sample value range corresponding to the user sample;

screening out the learning samples of which the user samples are in the sample value range from the plurality of groups of learning samples, and outputting the screened learning samples to a learning sample set;

and processing the learning sample set through the processing algorithm.

3. The method of claim 2, wherein after outputting the screened learning samples to a learning sample set, further comprising:

detecting whether repeated samples exist in the learning sample set;

if the repeated samples are detected, only one group of the learning samples in the repeated samples is reserved in the learning sample set.

4. The method of claim 1, wherein if there is an upper limit on the number of additional service sets and there are multiple secondary service types, the obtaining the secondary service type in the service selection model and adding the secondary service type to the additional service sets comprises:

Obtaining a plurality of gain values of the sample node obtained through the processing algorithm, wherein the gain values are in one-to-one correlation with a plurality of secondary service types;

sorting the plurality of secondary service types according to the values of the plurality of gain values to generate a secondary service sequence;

and sequentially selecting the secondary service types from the secondary service sequence from front to back, and adding the secondary service types to the additional service set until the upper limit of the number of the additional service sets is reached.

5. A terminal device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the following steps when executing the computer program:

Wherein, the business selection model is constructed by the following method:

6. The terminal device according to claim 5, wherein the processing the plurality of sets of learning samples by a preset processing algorithm includes:

acquiring a preset sample value range corresponding to the user sample;

and processing the learning sample set through the processing algorithm.

7. The terminal device of claim 6, wherein after outputting the screened learning samples to a learning sample set, further comprising:

detecting whether repeated samples exist in the learning sample set;

8. A computer readable storage medium storing a computer program, characterized in that the computer program when executed by a processor implements the steps of the method according to any one of claims 1 to 4.