CN115496158A

CN115496158A - Object value prediction method, device, computer equipment and storage medium

Info

Publication number: CN115496158A
Application number: CN202211199417.6A
Authority: CN
Inventors: 范伟杰; 王保山
Original assignee: Shanghai Pudong Development Bank Co Ltd
Current assignee: Shanghai Pudong Development Bank Co Ltd
Priority date: 2022-09-29
Filing date: 2022-09-29
Publication date: 2022-12-20

Abstract

The application relates to an object value prediction method, an object value prediction device, a computer device, a storage medium and a computer program product. The method comprises the following steps: clustering object features of all objects to be predicted in an object set to be predicted, and determining a first prediction value based on resource residual peak values of all objects to be predicted in each object group; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value; determining a third prediction value for each object to be predicted in the object set to be predicted based on the multi-dimensional relation data between the current object to be predicted and each first object; determining a four-prediction value based on the resource residual peak value of the current object to be predicted; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value. By adopting the method, the object value prediction accuracy can be improved.

Description

Object value prediction method, device, computer equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a method, an apparatus, a computer device, a storage medium, and a computer program product for predicting an object value.

Background

Most modern enterprises have a large number of business objects, and hierarchical interaction is carried out on the business objects with different values, so that the method has an important effect on improving the overall business level of the enterprises. In actual business, the activity of a large part of business objects is low, and the value of the business objects cannot be directly obtained, so that value prediction is needed.

In the existing object value prediction method, the value index of an object is usually formed manually according to a business expert rule. However, the object value prediction method can only classify the objects according to the data already shown by the client, and cannot predict the potential values of the objects, so that the object value prediction accuracy is low.

Disclosure of Invention

Based on this, it is necessary to provide an object value prediction method, an apparatus, a computer device, a computer-readable storage medium, and a computer program product capable of improving the accuracy of object value prediction, in view of the problem that the accuracy of the conventional object value prediction method is low.

In a first aspect, the present application provides a method for predicting an object value. The method comprises the following steps:

acquiring object characteristics of each object to be predicted in an object set to be predicted;

clustering object features of all objects to be predicted, dividing all the objects to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on a resource residual peak value of each object to be predicted in each object group in a first preset time period;

respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted;

determining multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted aiming at each object to be predicted in the object set to be predicted; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data;

determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period for each object to be predicted in the set of objects to be predicted;

and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted.

In one embodiment, the training step of the regression model includes:

acquiring object features of a training object and a value label of the training object;

inputting the object characteristics of the training object into the current training model to obtain the prediction value of the training object, judging whether the current training model meets the preset stop condition or not based on the prediction value and the value label, if so, taking the current training model as a regression model, and if not, continuing training the current training model until the current training model meets the preset stop condition.

In one embodiment, determining a third prediction value of the current object to be predicted based on the multidimensional relation data includes:

determining a sub-graph spectrum corresponding to the current object to be predicted on a knowledge graph of the object set to be predicted;

for each first object except the current object to be predicted in the object set to be predicted, carrying out weighted average on multi-dimensional relation data of the current object to be predicted and the current first object to obtain the intimacy distance between the current object to be predicted and the current first object;

obtaining a propagation weight between the current object to be predicted and the current first object according to the intimacy distance between the current object to be predicted and the current first object;

and determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weight between the current object to be predicted and each first object.

In one embodiment, determining a sub-graph spectrum corresponding to a current object to be predicted on a knowledge graph of a set of objects to be predicted includes:

constructing a knowledge graph of an object set to be predicted, wherein each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph;

selecting relation nodes in a preset series relation range of the current object to be predicted from a knowledge graph spectrum by taking the nodes corresponding to the current object to be predicted as a center;

and determining a sub-graph spectrum corresponding to the current object to be predicted based on the node corresponding to the current object to be predicted and the relation node.

In one embodiment, determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weights between the current object to be predicted and the first objects includes:

determining the shortest path from each relationship node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes in the sub-graph spectrum corresponding to the current object to be predicted;

acquiring the propagation weight between two adjacent target nodes in the target nodes;

acquiring a resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period;

determining the value influence of the object to be predicted corresponding to each Guan Jijie point on the current object to be predicted based on the propagation weight between two adjacent target nodes in the target nodes and the resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period;

and summing the value influence of the object to be predicted corresponding to each relation node on the current object to be predicted, and determining a third prediction value of the current object to be predicted.

In one embodiment, determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the set of objects to be predicted comprises:

and selecting the maximum value from the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted according to each object to be predicted in the object set to be predicted, and taking the maximum value as the object value of the current object to be predicted.

In a second aspect, the application also provides an object value prediction device. The device comprises:

the characteristic acquisition module is used for acquiring the object characteristics of each object to be predicted in the object set to be predicted;

the first value obtaining module is used for clustering the object characteristics of each object to be predicted, dividing each object to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on the resource residual peak value of each object to be predicted in each object group in a first preset time period;

the second value obtaining module is used for respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted;

the third valence value acquisition module is used for determining multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted aiming at each object to be predicted in the object set to be predicted; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data;

the fourth-value obtaining module is used for determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period for each object to be predicted in the set of objects to be predicted;

and the value determining module is used for determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted.

In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:

In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:

In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:

According to the object value prediction method, the device, the computer equipment, the storage medium and the computer program product, a plurality of object groups are obtained by clustering the object characteristics of all objects to be predicted in the object group to be predicted, and the first prediction value of all the objects to be predicted in the same object group can be determined based on the resource residual peak value of all the objects to be predicted in each object group in a first preset time period; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model, and obtaining a second prediction value of each object to be predicted in the object set to be predicted through the regression model; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted; determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted. The method for predicting the object value through the clustering, the regression model, the multi-dimensional relation data and the resource residual value in multiple angles can avoid deviation caused by predicting the object value through a single angle, the prediction angle is more comprehensive, and the accuracy of the prediction result is improved.

Drawings

FIG. 1 is a diagram of an exemplary embodiment of an application environment for a method for predicting a value of an object;

FIG. 2 is a flow diagram illustrating a method for object value prediction in one embodiment;

FIG. 3 is a schematic diagram illustrating a sub-flow of S204 in one embodiment;

FIG. 4 is a schematic diagram illustrating a sub-flow of S302 according to an embodiment;

FIG. 5 is a schematic diagram illustrating a sub-flow of S308 according to an embodiment;

FIG. 6 is a flow chart illustrating a method for predicting a value of an object according to another embodiment;

FIG. 7 is a block diagram showing the structure of an object value prediction apparatus according to an embodiment;

FIG. 8 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The object value prediction method provided by the embodiment of the application can be applied to the application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The data storage system may store data that the server 104 needs to process. The data storage system may be integrated on the server 104, or may be located on the cloud or other network server. The terminal 102 acquires object characteristics of each object to be predicted in the object set to be predicted; clustering object features of all objects to be predicted, dividing all the objects to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on a resource residual peak value of each object to be predicted in each object group in a first preset time period; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted; determining multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted aiming at each object to be predicted in the object set to be predicted; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data; determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period for each object to be predicted in the set of objects to be predicted; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices, and portable wearable devices, and the internet of things devices may be smart speakers, smart televisions, smart air conditioners, smart car-mounted devices, and the like. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, and the like. The server 104 may be implemented as a stand-alone server or as a server cluster comprised of multiple servers.

In one embodiment, as shown in fig. 2, an object value prediction method is provided, which is described by taking the method as an example applied to the terminal 102 in fig. 1, and includes the following steps:

s201, acquiring object characteristics of each object to be predicted in the object set to be predicted.

The object set to be predicted comprises at least one object to be predicted, and the object to be predicted is an object needing value prediction. The object characteristics refer to characteristics of the object to be predicted, which are related to specific services, and the types of the object characteristics comprise any one of state type characteristics, product type characteristics or event type characteristics. Illustratively, the state class characteristics include, but are not limited to, any of basic information, account opening status, or business state of the object. Product class characteristics include, but are not limited to, any of resource transfer scenarios, resource holding quantities, or object ratings. Event-like features include, but are not limited to, any of resource transfer-in and transfer-out events, resource transfer streams, or log-in streams.

The method for acquiring the object characteristics of each object to be predicted in the object set to be predicted by the terminal comprises the steps that the terminal acquires service data of each object to be predicted in the object set to be predicted, and performs characteristic engineering processing on the service data to acquire the object characteristics of each object to be predicted in the object set to be predicted. The characteristic engineering refers to screening better data characteristics from the business data in a series of engineering modes so as to improve the training effect of the model. Feature engineering includes, but is not limited to, any of data preprocessing, feature selection, and dimension reduction. Illustratively, the feature engineering further comprises missing value supplement, wherein the missing value supplement refers to replacing null values of numerical type in the service data with preset numerical values, and replacing null values of character type or null character strings in the service data with preset characters. Illustratively, the feature engineering further comprises one-hot coding, wherein the one-hot coding refers to replacing a category variable in the business data with a plurality of variables, and a new feature is introduced into each category so as to represent any number of categories.

S202, clustering the object characteristics of the objects to be predicted, dividing the objects to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on the resource residual peak value of each object to be predicted in each object group in a first preset time period.

The clustering refers to a method for classifying object features of each object to be predicted by adopting a clustering algorithm. The clustering algorithm is based on similarity, and in the object groups of the objects to be predicted obtained through the clustering algorithm, the objects to be predicted in each object group have higher similarity. Common Clustering algorithms include, but are not limited to, kmeans (K-means), DBSCAN (Density-Based Spatial Clustering of Applications with Noise), spectral Clustering, laplace mapping, PCA (Principal Component Analysis), GMM (Gaussian Mixture Model), meanshift (mean shift), and hierarchical Clustering. Specifically, the terminal divides each object to be predicted into a plurality of object groups by clustering object features of each object to be predicted.

The resource residual peak value refers to a resource residual value with the largest value in the resource residual value distribution of the object to be predicted in a preset time period. The object value prediction is carried out based on the clustering algorithm, and the efficiency of the object value prediction can be improved. In a plurality of object groups determined based on a clustering algorithm, each object to be predicted in each object group has higher similarity, a terminal takes a resource residual peak value of each object to be predicted in each object group in a first preset time period as a first prediction value of each object to be predicted in each object group, the first prediction values of each object to be predicted in each object group are the same, the prediction value of each object to be predicted in each object group is favorably obtained, and the accuracy of object value prediction is improved.

S203, respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted.

The regression model is a machine learning model with supervised learning and is used for establishing a mapping relation between independent variables and observed values. Common regression models include, but are not limited to, any of linear regression, polynomial regression, ridge regression, lasso regression, elastic network regression, or XGBoost (eXtreme Gradient Boosting). The regression model is a well-trained machine learning model. Before the terminal respectively inputs the object characteristics of each object to be predicted in the object set to be predicted into the regression model, the terminal also needs to train the regression model until the prediction stopping condition is met, and the obtained machine learning model after training is the regression model. And the terminal respectively inputs the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted.

S204, aiming at each object to be predicted in the object set to be predicted, determining multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted; and determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data.

The first object refers to an object to be predicted except for the current object to be predicted in the object set to be predicted. The multidimensional relation data refers to relation data in multiple dimensions between an object to be predicted and a first object. The multidimensional relation data includes, but is not limited to, any one of resource transfer relation data, resource recommendation relation data, internet Protocol Address (IP Address) affinity, similarity of latitude and longitude of a login Address, or resource roll-out organization relation data. The multidimensional relation between the current object to be predicted and the first object is directional, and specifically, the multidimensional relation may be unidirectional or bidirectional. And for each object to be predicted in the object set to be predicted, the terminal determines the multidimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted, and determines the third prediction value of the current object to be predicted according to the multidimensional relation data between the current object to be predicted and the first object and the knowledge graph of the object set to be predicted. The knowledge graph of the object set to be predicted is a mesh knowledge structure reflecting the relation of each object to be predicted in the object set to be predicted.

S205, aiming at each object to be predicted in the object set to be predicted, determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period.

And the terminal acquires the resource residual peak value of the current object to be predicted in a second preset time period aiming at each object to be predicted in the set of objects to be predicted, and takes the resource residual peak value of the current object to be predicted in the second preset time period as the fourth prediction value of the current object to be predicted.

S206, determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted.

The terminal fuses the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted, and the fusion result is used as the object value of each object to be predicted. The terminal fuses the prediction value results obtained by the four methods, the determined object value can simultaneously meet the prediction conditions of the four angles, and the object value has high accuracy.

In the object value prediction method, a plurality of object groups are obtained by clustering the object features of all objects to be predicted in an object set to be predicted, and the first prediction value of all the objects to be predicted in the same object group can be determined based on the resource residual peak value of all the objects to be predicted in each object group in a first preset time period; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model, and obtaining a second prediction value of each object to be predicted in the object set to be predicted through the regression model; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted; determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted. The method for predicting the object value through the clustering, the regression model, the multi-dimensional relation data and the resource residual value in multiple angles can avoid deviation caused by predicting the object value through a single angle, the prediction angle is more comprehensive, and the accuracy of the prediction result is improved.

In one embodiment, the training step of the regression model includes: acquiring object characteristics of a training object and a value label of the training object; inputting the object characteristics of the training object into the current training model to obtain the prediction value of the training object, judging whether the current training model meets the preset stop condition or not based on the prediction value and the value label, if so, taking the current training model as a regression model, and if not, continuing training the current training model until the current training model meets the preset stop condition.

Wherein the training object is an object of which the value is known. The value label is label information used to characterize the value of the object of the training subject. The current training model refers to a regression model before model training, and the regression model is obtained by the current training model through model training.

And the terminal obtains the object characteristics of the training object and the value label of the training object, and inputs the object characteristics of the training object into the current training model to obtain the prediction value of the training object. And the terminal compares the predicted value of the training object with the value label to obtain a loss function of the current training model. Judging whether the loss function is within a preset loss range, if so, enabling the current training model to meet preset conditions, enabling the current training model to serve as a regression model by the terminal, and if not, enabling the current training model not to meet the preset conditions, and continuing training the current training model by the terminal until the loss function is within the preset range and the current training model meets preset stop conditions. The loss function includes, but is not limited to, any of RMSE (Root Mean Square Error), MSE (Mean Square Error), MAE (Mean Absolute Error), SD (Standard Deviation), or R2 (R-Square).

In the embodiment, the object characteristics of the training object and the value label of the training object are obtained, the object characteristics of the training object are input into the current training model to obtain the prediction value of the training object, whether the current training model meets the preset stop condition is judged based on the prediction value and the value label, model training is performed based on the value label of the training object, and the regression model obtained through training can be guaranteed to have high object value prediction accuracy.

In one embodiment, as shown in fig. 3, determining a third prediction value of the current object to be predicted based on the multidimensional relation data includes:

s302, determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted.

The knowledge graph is a knowledge structure describing resource transfer nodes through which the resource transfer data flow and the sequence of the resource transfer data flow through the resource transfer nodes in a symbol form. The knowledge graph of the object set to be predicted comprises each object to be predicted in the object set to be predicted. The sub-map is a map formed by nodes corresponding to the current object to be predicted and nodes directly or indirectly connected with the current object to be predicted on the knowledge map. And the terminal determines a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted.

S304, for each first object except the current object to be predicted in the object set to be predicted, carrying out weighted average on the multidimensional relation data of the current object to be predicted and the current first object to obtain the intimacy distance between the current object to be predicted and the current first object.

And for each first object except the current object to be predicted in the object set to be predicted, the terminal performs weighted average on the multidimensional relation data of the current object to be predicted and the current first object to obtain the intimacy distance between the current object to be predicted and the current first object. In some embodiments, the multidimensional relationship data may be affinity in multiple dimensions, the affinity ranging between 0 and 1. Closer affinity to 0 indicates that the two objects are more intimate in the current dimensional relationship. Closer affinity to 1 indicates that the two objects are further apart in the current dimensional relationship. Illustratively, the intimacy between the current object to be predicted and the current first object under the dimension k is defined as

Wherein

The intimacy distance between the current object to be predicted and the current first object can be represented as:

s306, according to the intimacy distance between the current object to be predicted and the current first object, the propagation weight between the current object to be predicted and the current first object is obtained.

The propagation weight refers to the weight for resource transfer between the current object to be predicted and the current first object, and the greater the intimacy distance, the smaller the propagation weight, and the smaller the intimacy distance, the greater the propagation weight. And the terminal takes the difference value between the intimacy distance between the current object to be predicted and the current first object and 1 as the propagation weight between the current object to be predicted and the current first object.

S308, determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weight between the current object to be predicted and each first object.

The terminal obtains the propagation weight between the current object to be predicted and each first object. In the sub-graph spectrum corresponding to the current object to be predicted, each first object only comprises the object directly or indirectly related to the current object to be predicted. And determining a third prediction value of the current object to be predicted by combining the propagation weight between the current object to be predicted and the object directly or indirectly related to the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted.

In the embodiment, a sub-graph spectrum corresponding to the current object to be predicted is determined on a knowledge graph of the object set to be predicted, and for each first object except the current object to be predicted in the object set to be predicted, multi-dimensional relation data of the current object to be predicted and the current first object are subjected to weighted average to obtain the intimacy distance between the current object to be predicted and the current first object and propagation weight.

In one embodiment, as shown in fig. 4, determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted includes:

s402, constructing a knowledge graph of the object set to be predicted, wherein each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph.

The terminal constructs a knowledge graph of an object set to be predicted. The basic composition units of the knowledge graph are entity-relation-entity triples, entities and related attributes thereof, and the entities are connected with each other through relations to form the knowledge graph. And each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph of the object set to be predicted one by one. The relationship between each object to be predicted forms the relationship between the entities in the knowledge graph of the object set to be predicted.

S404, taking the node corresponding to the current object to be predicted as a center, and selecting a relation node in a preset series relation range of the current object to be predicted in the knowledge graph spectrum.

The relation node is arranged on the object to be predicted which is directly or indirectly connected with the current object to be predicted. If the current object to be predicted is directly connected with the relationship node, the relationship node is called as the relationship node in the first-level number relationship range of the object to be predicted. And if the current object to be predicted is indirectly connected with the relationship node through one node, the relationship node is called as the relationship node in the second-level relationship range of the object to be predicted. Correspondingly, if the current object to be predicted is indirectly connected with the relationship node through N nodes, the relationship node is called as the relationship node in the Nth series relationship range of the object to be predicted. And the terminal selects the relation nodes in the preset series relation range of the current object to be predicted from the knowledge graph spectrum by taking the nodes corresponding to the current object to be predicted as the center.

S406, determining a sub-graph spectrum corresponding to the current object to be predicted based on the node and the relation node corresponding to the current object to be predicted.

The terminal takes the node and the relation node corresponding to the current object to be predicted as the entity in the sub-graph spectrum corresponding to the current object to be predicted, and takes the connection relation between the node and the relation node corresponding to the current object to be predicted as the relation between the entity and the entity in the sub-graph spectrum, so that the sub-graph spectrum corresponding to the current object to be predicted is determined.

In this embodiment, a knowledge graph of an object set to be predicted is constructed, and a relationship node in a preset series relationship range of a current object to be predicted is selected from the knowledge graph spectrum by taking a node corresponding to the current object to be predicted as a center, so as to determine a sub-graph spectrum corresponding to the current object to be predicted. The sub-graph spectrum corresponding to the current object to be predicted is determined by the node corresponding to the current object to be predicted, the relation node and the connection relation between the node corresponding to the current object to be predicted and the relation node, the value of the current object to be predicted is predicted in the sub-graph spectrum corresponding to the current object to be predicted, and the accuracy of the value prediction of the object can be improved.

In one embodiment, as shown in fig. 5, determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weights between the current object to be predicted and the respective first objects includes:

s501, in the sub-graph spectrum corresponding to the current object to be predicted, the shortest path from each relation node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes are determined.

In the sub-graph corresponding to the current object to be predicted, each relationship node can directly or indirectly reach the node corresponding to the current object to be predicted. In the sub-graph spectrum corresponding to the current object to be predicted, the terminal determines the shortest path from each relationship node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes. In some embodiments, only one path from each relationship node to the node corresponding to the current object to be predicted is provided, and the terminal determines the unique path and the target node through which the unique path passes. In other embodiments, each relationship node includes a plurality of paths to reach the node corresponding to the current object to be predicted, and the terminal selects a shortest path from the plurality of paths as the shortest path from each relationship node to the node corresponding to the current object to be predicted, and obtains the target node through which the shortest path passes.

S502, acquiring the propagation weight between two adjacent target nodes in the target nodes.

The terminal obtains the propagation weight between two adjacent target nodes in the target nodes. Namely, the terminal obtains the propagation weight between the objects to be predicted corresponding to two adjacent target nodes in the target nodes.

S503, acquiring the resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in the third preset time period.

The terminal obtains a resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period.

S504, based on the propagation weight between two adjacent target nodes in the target nodes and the resource residual peak value of the object to be predicted corresponding to each Guan Jijie in the third preset time period, determining the value influence of the object to be predicted corresponding to each Guan Jijie on the current object to be predicted.

The terminal multiplies the propagation weight between two adjacent target nodes in the target nodes, determines a multiplication result, multiplies the multiplication result by a resource residual peak value of an object to be predicted corresponding to each Guan Jijie point in a third preset time period, and determines the value influence of the object to be predicted corresponding to each Guan Jijie point on the current object to be predicted.

And S505, summing the value influence of the object to be predicted corresponding to each relation node on the current object to be predicted, and determining a third prediction value of the current object to be predicted.

The terminal sums the value influence of the object to be predicted corresponding to each relation node on the current object to be predicted to obtain a summation result, and the summation result is used as a third prediction value of the current object to be predicted. In some embodiments, P is defined _i For the node corresponding to the current object to be predicted, each relation node p _j The resource residual peak value of the corresponding object to be predicted in the third preset time period is alpha _j ，p _j To P _i Has a shortest path of j-k ₁ —k ₂ —…—k _l I, then the node p _j To node P _i The value impact of (A) is:

wherein k is ₀ ＝j、k _l+1 = i, node P _i May be expressed as:

in this embodiment, the shortest path from each relationship node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes are determined in the sub-graph spectrum corresponding to the current object to be predicted. And determining the value influence corresponding to each Guan Jijie point based on the propagation weight and the resource residual peak value, wherein the third prediction value of the current object to be predicted is the sum of the value influences corresponding to each relation node. The method for determining the third prediction value by the sum of the value influence of each relation node on the current object to be predicted in the sub-map corresponding to the current object to be predicted can improve the accuracy of object value prediction.

In one embodiment, determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted comprises: and selecting the maximum value from the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted according to each object to be predicted in the object set to be predicted, and taking the maximum value as the object value of the current object to be predicted.

The terminal selects a maximum value from a first prediction value, a second prediction value, a third prediction value and a fourth prediction value of the current object to be predicted according to each object to be predicted in the object set to be predicted, and the maximum value is used as the object value of the current object to be predicted. In some embodiments, the terminal may further select any one of the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted as the object value of the current object to be predicted. In other embodiments, the terminal may further use any number of values of the first prediction value, the second prediction value, the third prediction value, and the fourth prediction value of the current object to be predicted, and use an average value of the values as the object value of the current object to be predicted.

In the embodiment, the maximum value of the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted is used as the object value of the current object to be predicted, so that deviation caused by single-angle prediction of the object value can be avoided, the prediction angle is more comprehensive, and the accuracy of the prediction result is improved.

To explain the object value prediction method and effect in the present solution in detail, the following description is made with a most detailed embodiment:

the method aims at the application field of object value prediction in the financial field. Fig. 6 is a flow chart of the object value prediction method. The method comprises the steps that a terminal obtains service data of each object to be predicted in an object set to be predicted, characteristic engineering processing is conducted on the service data, and object characteristics of each object to be predicted in the object set to be predicted are obtained.

The method comprises the steps of clustering object features of objects to be predicted, dividing the objects to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on a resource residual peak value of each object to be predicted in each object group in a first preset time period. The clustering algorithm used includes, but is not limited to, mini-batch Kmeans. And evaluating the clustering algorithm by adopting an MSE loss function.

And respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted. The regression model employed includes, but is not limited to, the Xgboost regression model. The training step of the regression model comprises the following steps: acquiring object features of a training object and a value label of the training object; inputting the object characteristics of the training object into the current training model to obtain the prediction value of the training object, judging whether the current training model meets the preset stop condition or not based on the prediction value and the value label, if so, taking the current training model as a regression model, and if not, continuing training the current training model until the current training model meets the preset stop condition. Wherein the training object is an object of which the value is known. Value labels are label information used to characterize the value of an object of a training object. The current training model refers to a regression model before model training, and the current training model is trained through the model to obtain the regression model.

And determining multidimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted according to each object to be predicted in the object set to be predicted, and determining a third prediction value of the current object to be predicted based on the multidimensional relation data. The multidimensional relation data refers to relation data in multiple dimensions between an object to be predicted and a first object. The multidimensional relation data comprises any one of but not limited to resource transfer relation data, resource recommendation relation data, IP address intimacy, log-in address longitude and latitude similarity or resource transfer-out organization relation data.

Determining a third prediction value of the current object to be predicted based on the multidimensional relation data, wherein the third prediction value comprises the following steps: constructing a knowledge graph of an object set to be predicted, wherein each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph, taking the node corresponding to the current object to be predicted as a center, selecting a relation node in a preset series relation range of the current object to be predicted in the knowledge graph, determining a sub-graph spectrum corresponding to the current object to be predicted based on the node and the relation node corresponding to the current object to be predicted, and determining the sub-graph spectrum

Can be expressed as:

wherein the node P _i For the node corresponding to the current object to be predicted, node p _j Is a relationship node, mind _ji Representing a node p _j To node P _i Distance of shortest path, node P _i The m-order neighbor subgraph is the subgraph spectrum corresponding to the current object to be predicted.

For each first object except the current object to be predicted in the object set to be predicted, carrying out weighted average on the multidimensional relation data of the current object to be predicted and the current first object to obtain the current object to be predicted and the current first objectThe intimacy distance of (a). The multidimensional relationship data may be intimacy in multiple dimensions, the intimacy ranging from 0 to 1. Closer affinity to 0 indicates that the two objects are more intimate in the current dimensional relationship. Closer affinity to 1 indicates that the two objects are further apart in the current dimensional relationship. Illustratively, the intimacy between the current object to be predicted and the current first object under the dimension k is defined as

Wherein

The intimacy distance between the current object to be predicted and the current first object can be expressed as:

obtaining a propagation weight omega between the current object to be predicted and the current first object according to the intimacy distance between the current object to be predicted and the current first object _ij ：ω _ij ＝1-d _ij 。

And in the sub-graph spectrum corresponding to the current object to be predicted, determining the shortest path from each relationship node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes. And acquiring the propagation weight between two adjacent target nodes in the target nodes, and acquiring the resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period. Based on the propagation weight between two adjacent target nodes in the target nodes and the resource residual peak value of the object to be predicted corresponding to each Guan Jijie in the third preset time period, determining the value influence of the object to be predicted corresponding to each Guan Jijie on the current object to be predicted, summing the value influence of the object to be predicted corresponding to each relationship node on the current object to be predicted, and determining the third prediction value of the current object to be predicted. Definition P _i For the node corresponding to the current object to be predicted, each relation node p _j The resource residual peak value of the corresponding object to be predicted in the third preset time period isα _j ，p _j To P _i Is j-k as the shortest path ₁ —k ₂ —…—k _l I, then the node p _j To node P _i The value impact of (c) is:

wherein k is ₀ ＝j、k _l+1 = i, node P _i May be expressed as:

and for each object to be predicted in the object set to be predicted, the terminal acquires a resource residual peak value of the current object to be predicted in a second preset time period, and the resource residual peak value of the current object to be predicted in the second preset time period is used as a fourth prediction value of the current object to be predicted.

And for each object to be predicted in the object set to be predicted, the terminal selects the maximum value from the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted, and the maximum value is used as the object value of the current object to be predicted. In some embodiments, the terminal may further select any one of the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted as the object value of the current object to be predicted. In other embodiments, the terminal may further use any number of values of the first prediction value, the second prediction value, the third prediction value, and the fourth prediction value of the current object to be predicted, and use an average value of the values as the object value of the current object to be predicted.

The object value prediction method is characterized in that a plurality of object groups are obtained by clustering object features of all objects to be predicted in an object set to be predicted, and the first prediction value of all the objects to be predicted in the same object group can be determined based on the resource residual peak value of all the objects to be predicted in each object group in a first preset time period; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model, and obtaining a second prediction value of each object to be predicted in the object set to be predicted through the regression model; determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted; determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted. The method for predicting the object value through the clustering, the regression model, the multi-dimensional relation data and the resource residual value in multiple angles can avoid deviation caused by predicting the object value through a single angle, the prediction angle is more comprehensive, and the accuracy of the prediction result is improved.

It should be understood that, although the steps in the flowcharts related to the embodiments as described above are sequentially displayed as indicated by arrows, the steps are not necessarily performed sequentially as indicated by the arrows. The steps are not limited to being performed in the exact order illustrated and, unless explicitly stated herein, may be performed in other orders. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.

Based on the same inventive concept, the embodiment of the present application further provides an object value prediction apparatus for implementing the above-mentioned object value prediction method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the method, so the specific limitations in one or more embodiments of the object value prediction device provided below can be referred to the limitations of the object value prediction method in the above, and details are not repeated here.

In one embodiment, as shown in fig. 7, there is provided an object value prediction apparatus 100 including: a feature obtaining module 110, a first valence obtaining module 120, a second valence obtaining module 130, a third valence obtaining module 140, a fourth valence obtaining module 150, and a value determining module 160, wherein:

the feature obtaining module 110 is configured to obtain object features of each object to be predicted in the object set to be predicted.

The first value obtaining module 120 is configured to cluster object features of each object to be predicted, divide each object to be predicted into a plurality of object groups, and determine a first prediction value of each object to be predicted in each object group based on a resource remaining peak value of each object to be predicted in each object group within a first preset time period.

The second value obtaining module 130 is configured to input the object features of each object to be predicted in the object set to be predicted to the regression model, so as to obtain a second prediction value of each object to be predicted in the object set to be predicted.

A third valence value obtaining module 140, configured to determine, for each object to be predicted in the object set to be predicted, multidimensional relation data between the current object to be predicted and each first object in the object set to be predicted, except the current object to be predicted; and determining a third prediction value of the current object to be predicted based on the multi-dimensional relation data.

The fourth-value obtaining module 150 is configured to determine, for each object to be predicted in the set of objects to be predicted, a fourth prediction value of the current object to be predicted based on a resource remaining peak of the current object to be predicted in a second preset time period.

And the value determining module 160 is configured to determine the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the set of objects to be predicted.

The object value prediction device clusters the object characteristics of each object to be predicted in the object set to be predicted to obtain a plurality of object groups, and can determine the first prediction value of each object to be predicted in the same object group based on the resource residual peak value of each object to be predicted in each object group in a first preset time period; respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into the regression model, and obtaining a second prediction value of each object to be predicted in the object set to be predicted through the regression model; determining a third prediction value of the current object to be predicted based on multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted; determining a fourth prediction value of the current object to be predicted based on the resource residual peak value of the current object to be predicted in a second preset time period; and determining the object value of each object to be predicted based on the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of each object to be predicted in the object set to be predicted. The method for predicting the object value through the clustering, the regression model, the multi-dimensional relation data and the resource residual value in multiple angles can avoid deviation caused by predicting the object value through a single angle, the prediction angle is more comprehensive, and the accuracy of the prediction result is improved.

In one embodiment, in terms of training of the regression model, the second value obtaining module 130 is further configured to: acquiring object characteristics of a training object and a value label of the training object; inputting the object characteristics of the training object into the current training model to obtain the prediction value of the training object, judging whether the current training model meets the preset stop condition or not based on the prediction value and the value label, if so, taking the current training model as a regression model, and if not, continuing training the current training model until the current training model meets the preset stop condition.

In one embodiment, in determining a third predicted value of the current object to be predicted based on the multidimensional relation data, the third valence obtaining module 140 is further configured to: determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted; for each first object except the current object to be predicted in the object set to be predicted, carrying out weighted average on multi-dimensional relation data of the current object to be predicted and the current first object to obtain the intimacy distance between the current object to be predicted and the current first object; acquiring a propagation weight between the current object to be predicted and the current first object according to the intimacy distance between the current object to be predicted and the current first object; and determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weight between the current object to be predicted and each first object.

In an embodiment, in determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted, the third valence obtaining module 140 is further configured to: constructing a knowledge graph of an object set to be predicted, wherein each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph; selecting relation nodes in a preset series relation range of the current object to be predicted from a knowledge graph spectrum by taking the nodes corresponding to the current object to be predicted as a center; and determining a sub-graph spectrum corresponding to the current object to be predicted based on the node corresponding to the current object to be predicted and the relation node.

In one embodiment, in determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weight between the current object to be predicted and each first object, the third value obtaining module 140 is further configured to: determining the shortest path from each relationship node to the node corresponding to the current object to be predicted and the target node through which the shortest path passes in the sub-graph spectrum corresponding to the current object to be predicted; acquiring the propagation weight between two adjacent target nodes in the target nodes; acquiring a resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period; determining the value influence of the object to be predicted corresponding to each Guan Jijie point on the current object to be predicted based on the propagation weight between two adjacent target nodes in the target nodes and the resource residual peak value of the object to be predicted corresponding to each Guan Jijie point in a third preset time period; and summing the value influence of the object to be predicted corresponding to each relation node on the current object to be predicted, and determining a third prediction value of the current object to be predicted.

In one embodiment, in determining the object value of each object to be predicted based on the first predicted value, the second predicted value, the third predicted value and the fourth predicted value of each object to be predicted in the set of objects to be predicted, the value determination module 160 is further configured to: and selecting the maximum value from the first prediction value, the second prediction value, the third prediction value and the fourth prediction value of the current object to be predicted according to each object to be predicted in the object set to be predicted, and taking the maximum value as the object value of the current object to be predicted.

The respective modules in the object value prediction apparatus may be wholly or partially implemented by software, hardware, or a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, an Input/Output interface (I/O for short), and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used for storing a set of objects to be predicted, object features, object groups, resource residual peaks, regression models, multidimensional relation data, a first prediction value, a second prediction value, a third prediction value and a fourth prediction value. The input/output interface of the computer device is used for exchanging information between the processor and an external device. The communication interface of the computer device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement a method of object value prediction.

It will be appreciated by those skilled in the art that the configuration shown in fig. 8 is a block diagram of only a portion of the configuration associated with the present application, and is not intended to limit the computing device to which the present application may be applied, and that a particular computing device may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In an embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored therein a computer program, which when executed by the processor, performs the steps of the above-described method embodiments.

In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

In an embodiment, a computer program product is provided, comprising a computer program which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

It should be noted that, the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use and processing of the related data need to comply with the relevant laws and regulations and standards of the relevant country and region.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, databases, or other media used in the embodiments provided herein can include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), magnetic Random Access Memory (MRAM), ferroelectric Random Access Memory (FRAM), phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases involved in the embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing based data processing logic devices, etc., without limitation.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method for predicting a value of an object, the method comprising:

respectively inputting the object characteristics of each object to be predicted in the object set to be predicted into a regression model to obtain a second prediction value of each object to be predicted in the object set to be predicted;

determining multi-dimensional relation data between a current object to be predicted and each first object except the current object to be predicted in an object set to be predicted, aiming at each object to be predicted in the object set to be predicted; determining a third prediction value of the current object to be predicted based on the multidimensional relation data;

determining a fourth prediction value of the current object to be predicted based on a resource residual peak value of the current object to be predicted in a second preset time period for each object to be predicted in the set of objects to be predicted;

2. The method of claim 1, wherein the step of training the regression model comprises:

inputting the object characteristics of the training object into a current training model to obtain the predicted value of the training object, judging whether the current training model meets a preset stopping condition or not based on the predicted value and the value label, if so, taking the current training model as the regression model, and if not, continuing training the current training model until the current training model meets the preset stopping condition.

3. The method of claim 1, wherein determining a third predicted value of the current object to be predicted based on the multidimensional relationship data comprises:

determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge graph of the object set to be predicted;

4. The method of claim 3, wherein the determining a sub-graph spectrum corresponding to the current object to be predicted on the knowledge-graph of the set of objects to be predicted comprises:

constructing a knowledge graph of the object set to be predicted, wherein each object to be predicted in the object set to be predicted corresponds to one node in the knowledge graph;

selecting a relation node in a preset series relation range of the current object to be predicted from the knowledge graph by taking a node corresponding to the current object to be predicted as a center;

5. The method according to claim 4, wherein the determining a third prediction value of the current object to be predicted in the sub-graph spectrum corresponding to the current object to be predicted based on the propagation weight between the current object to be predicted and each first object comprises:

determining the shortest path from each relationship node to the node corresponding to the current object to be predicted and a target node through which the shortest path passes in the sub-graph spectrum corresponding to the current object to be predicted;

6. The method of claim 1, wherein determining the object value of each object to be predicted based on the first predicted value, the second predicted value, the third predicted value and the fourth predicted value of each object to be predicted in the set of objects to be predicted comprises:

and aiming at each object to be predicted in the object set to be predicted, selecting a maximum value from a first prediction value, a second prediction value, a third prediction value and a fourth prediction value of the current object to be predicted, and taking the maximum value as the object value of the current object to be predicted.

7. An object value prediction apparatus, characterized in that the apparatus comprises:

the first value acquisition module is used for clustering the object characteristics of each object to be predicted, dividing each object to be predicted into a plurality of object groups, and determining a first prediction value of each object to be predicted in each object group based on the resource residual peak value of each object to be predicted in each object group in a first preset time period;

the third valence value acquisition module is used for determining multi-dimensional relation data between the current object to be predicted and each first object except the current object to be predicted in the object set to be predicted aiming at each object to be predicted in the object set to be predicted; determining a third prediction value of the current object to be predicted based on the multidimensional relation data;

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.

9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 6 when executed by a processor.