CN111105265A

CN111105265A - Prediction method and device based on customer information, computer equipment and storage medium

Info

Publication number: CN111105265A
Application number: CN201910969447.2A
Authority: CN
Inventors: 徐绪波
Original assignee: Ping An Bank Co Ltd
Current assignee: Ping An Bank Co Ltd
Priority date: 2019-10-12
Filing date: 2019-10-12
Publication date: 2020-05-05

Abstract

The invention discloses a client information-based prediction method, a client information-based prediction device, computer equipment and a storage medium. The method comprises the following steps: the method comprises the steps of converting historical customer information in a customer information base to obtain corresponding variable information, constructing and obtaining a plurality of probability prediction models corresponding to each preset prediction template according to the variable information, combining the probability prediction models according to a combination formula to obtain a combined probability prediction model, converting the customer information to be predicted to obtain customer variable information to be predicted, inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value, and obtaining a customer with a probability value larger than a positive sample proportion value of the historical customer information in the customer information to be predicted as a target customer. The invention is based on the prediction model technology, and based on the combined probability prediction model obtained by combining a plurality of probability prediction models, the efficiency and the accuracy of obtaining the target customer based on customer information prediction can be greatly improved.

Description

Prediction method and device based on customer information, computer equipment and storage medium

Technical Field

The present invention relates to the field of computer technologies, and in particular, to a prediction method and apparatus based on client information, a computer device, and a storage medium.

Background

The consumption habits of customers are usually directed to a certain result, the same consumption habits of different customers are usually directed to the same type of result, whether the customers have tendencies in a specific behavior can be predicted according to the consumption habits of the customers, the customers with corresponding behavior tendencies are determined as target customers, and if the target customers are the customers with the tendency to lose, prompt information can be sent to the target customers to save the customers before the customers lose. Therefore, the method in the prior art has the problem of low accuracy when the target customer is obtained by prediction based on the customer information.

Disclosure of Invention

The embodiment of the invention provides a client information-based prediction method, a client information-based prediction device, computer equipment and a storage medium, and aims to solve the problem of low accuracy in the prior art that the target client is obtained by prediction based on client information.

In a first aspect, an embodiment of the present invention provides a prediction method based on customer information, including:

converting historical customer information in a preset customer information base according to a preset information conversion model to obtain variable information corresponding to the historical customer information;

constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information;

combining a plurality of probability prediction models according to a preset combination formula to obtain a combined probability prediction model;

training the combined probability prediction model according to the variable information and a preset model training rule to obtain a trained combined probability prediction model;

if customer information to be predicted input by a user terminal is received, obtaining customer variable information to be predicted corresponding to the customer information to be predicted based on the information conversion model;

inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value;

and acquiring the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as a target client.

In a second aspect, an embodiment of the present invention provides a prediction apparatus based on customer information, including:

the client variable information acquisition unit is used for converting historical client information in a preset client information base according to a preset information conversion model to obtain variable information corresponding to the historical client information;

the combined probability prediction model construction unit is used for constructing and obtaining a plurality of probability prediction models corresponding to each preset prediction template according to the variable information;

the probability prediction model combination unit is used for combining a plurality of probability prediction models according to a preset combination formula to obtain a combined probability prediction model;

the model training unit is used for training the combined probability prediction model according to the variable information and a preset model training rule to obtain a trained combined probability prediction model;

the client variable information to be predicted acquiring unit is used for acquiring client variable information to be predicted corresponding to the client information to be predicted based on the information conversion model if the client information to be predicted input by the user terminal is received;

a probability value obtaining unit, configured to input the client variable information to be predicted into a trained combined probability prediction model to obtain a corresponding probability value;

and the target client obtaining unit is used for obtaining the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as the target client.

In a third aspect, an embodiment of the present invention further provides a computer device, which includes a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the processor implements the client information-based prediction method described in the first aspect.

In a fourth aspect, the present invention further provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and the computer program, when executed by a processor, causes the processor to execute the prediction method based on customer information according to the first aspect.

The embodiment of the invention provides a client information-based prediction method and device, computer equipment and a storage medium. The method comprises the steps of converting historical customer information in a customer information base to obtain corresponding variable information, constructing and obtaining a plurality of probability prediction models corresponding to each preset prediction template according to the variable information, combining the probability prediction models according to a combination formula to obtain a combined probability prediction model, converting the customer information to be predicted to obtain customer variable information to be predicted, inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value, and obtaining a customer with a probability value larger than a positive sample proportion value of the historical customer information in the customer information to be predicted as a target customer. By the method, the combined probability prediction model obtained based on the combination of the probability prediction models can greatly improve the efficiency and accuracy of obtaining the target customer based on customer information prediction.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

FIG. 1 is a flow chart illustrating a prediction method based on customer information according to an embodiment of the present invention;

FIG. 2 is a sub-flowchart of a prediction method based on customer information according to an embodiment of the present invention;

FIG. 3 is a schematic sub-flow chart of a prediction method based on customer information according to an embodiment of the present invention;

FIG. 4 is a schematic sub-flow chart of a prediction method based on customer information according to an embodiment of the present invention;

FIG. 5 is a schematic sub-flow chart of a prediction method based on customer information according to an embodiment of the present invention;

FIG. 6 is a schematic flow chart illustrating a prediction method based on customer information according to an embodiment of the present invention;

FIG. 7 is a schematic block diagram of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 8 is a schematic block diagram of a sub-unit of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 9 is a schematic block diagram of another sub-unit of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 10 is a schematic block diagram of another sub-unit of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 11 is a schematic block diagram of another sub-unit of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 12 is another schematic block diagram of a prediction apparatus based on customer information according to an embodiment of the present invention;

FIG. 13 is a schematic block diagram of a computer device provided by an embodiment of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It will be understood that the terms "comprises" and/or "comprising," when used in this specification and the appended claims, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

It is also to be understood that the terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in the specification of the present invention and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

It should be further understood that the term "and/or" as used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

Referring to fig. 1, fig. 1 is a flowchart illustrating a prediction method based on customer information according to an embodiment of the present invention. The prediction method based on the client information is applied to a user terminal, and the method is executed through application software installed in the user terminal, wherein the user terminal is a terminal device, such as a desktop computer, a notebook computer, a tablet computer or a mobile phone, which is used for executing the prediction method based on the client information so as to predict clients having tendencies in a specific behavior and screen target clients.

As shown in fig. 1, the method includes steps S110 to S170.

S110, historical customer information in a preset customer information base is converted according to a preset information conversion model to obtain variable information corresponding to the historical customer information.

And converting the historical client information in a preset client information base according to a preset information conversion model to obtain variable information corresponding to the historical client information. The customer information base comprises historical customer information corresponding to a plurality of customers, and the historical customer information comprises personal information and historical transaction information of the customers, for example, the historical customer information comprises personal information such as account numbers, ages, sexes and professions, and historical transaction information such as total assets, purchased product amount, transaction times, transaction amount and latest transaction time. The information conversion model is a model for performing normalization processing on historical customer information to convert the historical customer information into interval numerical values, and specifically comprises a customer rating rule, a sample judgment rule, a variable screening rule and a variable conversion rule.

For example, part of the historical customer information in the customer information base is shown in table 1.

TABLE 1

In an embodiment, as shown in fig. 2, step S110 includes sub-steps S111, S112 and S113.

And S111, judging whether each piece of historical customer information is a positive sample according to the sample judgment rule so as to obtain sample information corresponding to each piece of historical customer information.

And judging whether each piece of historical customer information is a positive sample according to the sample judgment rule so as to obtain customer sample information corresponding to each piece of historical customer information. The client information base comprises a plurality of clients, whether historical client information is a positive sample or not can be judged based on a sample judgment rule, for example, if the client asset probability value is analyzed, the client assets can be rapidly reduced to be regarded as the asset loss of the client as the assets have different asset variation degrees and activity degrees among the clients, so that the sample judgment rule can judge that the assets of the client in the last two months are continuously reduced by more than 20%, if the judgment result is yes, the sample information of the client is obtained as the positive sample, and if the judgment result is not, the sample information of the client is obtained as the negative sample; the sample judgment rule can also judge whether the client is inactive for a long time, the client which is inactive for a long time can be regarded as a lost client of the enterprise, if so, the sample information of the client is obtained as a positive sample, and if not, the sample information of the client is obtained as a negative sample.

For example, the sample judgment rule is to judge that the sample information of the customer is a positive sample if the purchased product in the historical customer information is "none" and the last transaction time is more than 180 days away from the current time. Then, according to the sample judgment rule, the client with the account number of S003 is judged to be a positive sample, and the clients with the account numbers of S001, S002 and S004 are judged to be negative samples.

And S112, acquiring a customer grade corresponding to each historical customer information in the customer information base according to the customer rating rule.

And acquiring the customer grade corresponding to each historical customer information in the customer information base according to the customer rating rule. The client rating rule is rule information used for acquiring the client level corresponding to each piece of historical client information, and an enterprise can classify clients to corresponding client levels according to self business requirements for conveniently managing the clients.

For example, the customer rating rule includes four customer levels of a small customer (total asset amount + total purchased product amount is less than or equal to 20000), a common customer (20000 < total asset amount + total purchased product amount is less than or equal to 100000), a large customer (100000 < total asset amount + total purchased product amount is less than or equal to 500000), and a honored guest customer (500000 < total asset amount + total purchased product amount), so that the customer with the account number S003 is the small customer, the customer with the account number S002 is the common customer, the customer with the account number S004 is the large customer, and the customer with the account number S001 is the honored guest customer.

And S113, converting the item value corresponding to the conversion item in the variable conversion rule, the customer sample information and the customer grade in a customer information base according to the variable conversion rule to obtain variable information corresponding to each historical customer information.

And converting the item value corresponding to the conversion item in the variable conversion rule, the customer sample information and the customer grade in a customer information base according to the variable conversion rule to obtain variable information corresponding to each historical customer information. In order to quantify the specific information of each client in the client information base, the information of each client is required to be converted into variable information through a variable conversion rule, the variable conversion rule comprises a plurality of preset conversion items, each conversion item corresponds to one item value in one piece of historical client information, the historical client information is converted, namely the information of the client is subjected to normalization processing, and the information required to be converted comprises the item value corresponding to the conversion item in the variable conversion rule in the client information base, client sample information and client grade. Specifically, the variable conversion rule includes a rule for converting each item value, and each item value corresponding to a conversion item can be converted to obtain a variable value, where the range of each variable value is [0, 1 ].

Specifically, for the sex-based dichotomous conversion item, the item value corresponding to the conversion item is directly determined, and if the item value corresponding to the sex-based conversion item is male, the variable value is correspondingly obtained as 1; if the item value corresponding to the conversion item of sex is "woman", the correspondence variable value is "0". The variable conversion rule further includes an activation function and preset intermediate values corresponding to conversion items such as the total amount of the assets and the transaction times, and the rules for converting the conversion items such as the total amount of the assets and the transaction times are f (i) -10 × (i-z) ÷ z, wherein i is an item value corresponding to a certain conversion item, z is an intermediate value of the corresponding conversion item, and f (i) is a conversion value obtained through conversion. Inputting the conversion value obtained by calculation into an activation function, namely calculating a variable value corresponding to a conversion item, and acquiring a variable value obtained by converting a piece of historical customer information as the variable information of the historical customer information.

For example, the conversion item of the transaction number is preset to have an intermediate value of 60, and the activation function is f (x) ═ 1+ e^-x) 1, obtaining a conversion value f (i) × (103-60) ÷ 60 ═ 0.7167 of a conversion item corresponding to the transaction number of times of the account S001 according to a conversion formula, inputting the conversion value 0.7167 into the activation function, and finally obtaining a variable value 0.3281 of the conversion item in the account S001.

And S120, constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information.

And constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information. And combining the variable information obtained by one prediction template to construct a probability prediction model corresponding to the prediction template, and inputting the variable information of one client into each probability prediction model to calculate and obtain a predicted value corresponding to each probability prediction model. Specifically, the prediction templates may be a logistic regression prediction template, a support vector machine prediction template (svm), and an eXtreme Gradient boost prediction template (XGBoost), each prediction template includes a prediction function, and the corresponding prediction value may be obtained by calculating an input value input to the prediction function through the prediction function.

In an embodiment, as shown in fig. 3, step S120 includes substeps S121 and S122.

And S121, constructing and obtaining an input node according to the conversion item in the variable information.

And constructing and obtaining an input node according to the conversion project in the variable information. Because the information input into the probability prediction models corresponds to the same conversion item, an input node of the probability prediction model can be constructed according to the conversion item in the variable information, the variable information corresponding to the conversion item is input through the input node, and the input value obtained through calculation is input into the probability prediction model for calculation, so that the predicted value corresponding to the probability prediction model can be obtained.

For example, the prediction function in the logistic regression prediction template is h (x) ═ 1+ e^-x) 1, if the variable information includes five conversion items, the input value x corresponds to five input nodes, and the relationship between x and the five input nodes is: x is ω₁×A₁+ω₂×A₂+ω₃×A₃+ω₄×A₄+ω₅×A₅+ θ, wherein ω_iAnd theta is a parameter value in the relation_iI.e. the input value of the i-th input node. When the input node is constructed, the parameter values in the relational expression are random values.

And S122, constructing a plurality of probability prediction models according to each prediction template and the input nodes.

And constructing a plurality of probability prediction models according to each prediction template and the input nodes. And combining the constructed input node with a prediction function in a prediction template to obtain a probability prediction model corresponding to the prediction template, wherein the output value corresponding to the input node is the input value of the probability prediction model.

S130, combining the probability prediction models according to a preset combination formula to obtain a combined probability prediction model.

And combining the probability prediction models according to a preset combination formula to obtain a combined probability prediction model. The combined probability prediction model comprises a plurality of probability prediction models, each prediction template corresponds to one probability prediction model constructed in the combined probability prediction model, the combined formula comprises a weight value corresponding to each probability prediction model, a final probability value can be obtained by multiplying a predicted value obtained by each probability prediction model with the corresponding weight value and accumulating, the plurality of probability prediction models are combined according to the combined formula to obtain the combined probability prediction model, the situation that a prediction result is incorrect due to a single probability prediction model can be avoided, and the accuracy of the prediction result can be greatly improved. Specifically, each probability prediction model in the combination formula corresponds to a weight value, and the final probability value can be obtained by accumulating products obtained by multiplying each probability prediction model by the corresponding weight value according to the combination formula.

For example, the corresponding combination formula of the combined probabilistic predictive model can be expressed as: l ═ a₁×H₁+a₂×H₂+……+a_i×H_i(ii) a Wherein H_iTo combine the predicted values of the ith probabilistic predictive model contained in the probabilistic predictive model, a_iAnd the weighted value corresponding to the ith prediction model is L, namely the probability value output by the combined probability prediction model.

And S140, training the combined probability prediction model according to the variable information and a preset model training rule to obtain the trained combined probability prediction model.

And training the combined probability prediction model according to the variable information and a preset model training rule to obtain the trained combined probability prediction model. The model training rules comprise data splitting information and parameter value adjusting rules. Before the probability value is calculated by using the combined probability prediction model, the combined probability prediction model needs to be trained according to the obtained variable information so as to improve the accuracy of the combined probability prediction model.

In an embodiment, as shown in fig. 4, step S140 includes sub-steps S141, S142, S143, and S144.

And S141, splitting the variable information into data sets with corresponding quantity according to the data splitting information.

And splitting the variable information into a corresponding number of data sets according to the data splitting information. The splitting information comprises the specific number of the variable information for average splitting, and the variable information can be averagely split into a plurality of data sets according to the splitting information.

For example, the variable information includes 1000 pieces of information corresponding to the clients, and if the split information is 10, the information corresponding to 1000 pieces of clients is split into 10 data sets on average, and each data set includes 100 pieces of information corresponding to the clients.

And S142, performing multiple rounds of training on all the probability prediction models according to the parameter value adjustment rule and the plurality of data sets to obtain a training result of each probability prediction model in the combined probability prediction model, wherein the training results comprise the accuracy and the coverage rate corresponding to each probability prediction model in each round of training.

The training process is also called a grid search method, one data set in the data sets is sequentially selected as a training data set, the other data sets are used as test data sets, all probability prediction models are subjected to multi-round training by combining parameter adjustment rules, and a training result of each probability prediction model in the combined probability prediction model is obtained, wherein the training result comprises the accuracy and the coverage rate corresponding to each probability prediction model in each round of training. Specifically, if the total number of the data sets is k, k rounds of cross training are performed on each probability prediction model, when a first round of training is performed on a probability prediction model, a first data set is used as a test data set, the rest k-1 data sets are used as training data sets, information corresponding to each client in the first training data set is input into the probability prediction model, an output result with a predicted value larger than 50% is used as a forward result, the number of positive samples corresponding to the sample information of the client corresponding to the forward result in the training data set is counted, and calculating to obtain the accuracy Z of the current training data set when the probability prediction model is trained as S/V, wherein, S is the number of positive samples of the sample information of the client corresponding to the positive result in the training data set, and V is the number of historical client information contained in the training data set. The parameter adjusting rule comprises an accuracy threshold, a parameter adjusting direction and a parameter adjusting amplitude, the parameter adjusting direction comprises positive adjustment and negative adjustment, the parameter adjusting amplitude is a specific amplitude value to be adjusted, whether the accuracy of the current training data set during training of the probability prediction model is smaller than the accuracy threshold is judged, and if the judgment result is not smaller than the accuracy threshold, the parameter value in the current probability prediction model is adjusted according to the amplitude values in the positive adjustment and the parameter adjusting amplitude in the parameter adjusting direction; and if the judgment result is less than the preset value, adjusting the parameter value in the current probability prediction model according to the reverse adjustment in the parameter adjustment direction and the amplitude value in the parameter adjustment amplitude.

For example, the amplitude value in the parameter adjustment amplitude is 0.05, and the determination result indicates that the accuracy when the current training data set trains a probability prediction model is not less than the accuracy threshold, the current adjustment needs to be performed in the forward direction, and the current adjustment is multiplied by 1.05 on the basis of the parameter value original value in the probability prediction model to obtain a new parameter value.

The parameter values in the probability prediction model can be adjusted once by one training data set, the probability prediction model is trained by k training data sets to obtain a probability prediction model after the first round of training, and the test data set is input into the probability prediction model after the first round of training to calculate the corresponding accuracy and coverage rate, namely, the one-round training of the probability prediction model is completed. The method for calculating the accuracy of the probability prediction model through the test data set is the same as the method for calculating the accuracy of the current training data set when the probability prediction model is trained, and the coverage rate F of the test data set when the probability prediction model is tested is calculated and obtained, wherein S is the number of positive samples of sample information of a client corresponding to a forward result in the test data set, and U is the total number of the positive samples contained in the test data set.

S143, selecting a parameter value corresponding to the round of training with the highest sum of the accuracy and the coverage rate corresponding to the probability prediction model as a parameter value in the probability prediction model according to the training result.

And selecting a parameter value corresponding to the round of training with the highest sum of the accuracy and the coverage rate corresponding to the probability prediction model as a parameter value in the probability prediction model according to the training result. And performing multiple rounds of cross training on each probability prediction model, wherein each probability prediction model corresponds to the accuracy and the coverage rate obtained by multiple rounds of training, and the parameter value corresponding to the highest training round with the accuracy and the coverage rate corresponding to the probability prediction model is used as the parameter value in the probability prediction model, namely the optimal parameter value corresponding to each probability prediction model is obtained by training.

S144, training the combined probability prediction model according to the data set to obtain the trained combined probability prediction model.

And training the combined probability prediction model according to the data set to obtain the trained combined probability prediction model. Specifically, the data sets are sequentially input into the trained probability prediction models in the combined probability prediction model, and after the current data set is input into the combined probability prediction model, the accuracy and the coverage rate of each probability prediction model are obtained, and the weighted value corresponding to the probability prediction model is adjusted according to the sum of the accuracy and the coverage rate of each probability prediction model, so that the combined probability prediction model is trained to obtain the trained combined probability prediction model. Specifically, a data set is input into the combined probability prediction model, the accuracy and the coverage rate of each probability prediction model are obtained according to the steps, all the probability prediction models are ranked according to the sum of the accuracy and the coverage rate, the probability prediction model with the highest ranking is adjusted in the positive direction on the basis of the original weighted value, the probability prediction model with the lowest ranking is adjusted in the negative direction on the basis of the original weighted value, and the sum of the weighted values of all the probability prediction models is 1, so that one-time training of the combined probability prediction model can be completed.

For example, the combined probability prediction model comprises three probability prediction models, the sum of the accuracy and the coverage rate of the A probability prediction model is the largest, the weighted value of the A probability prediction model is increased by 0.03 on the basis of the original weighted value, and the sum of the accuracy and the coverage rate of the B probability prediction model is the smallest; the weighted value of the B probability prediction model is reduced by 0.03 on the basis of the original weighted value, and the sum of the accuracy rate and the coverage rate of the C probability prediction model is between the A probability prediction model and the B probability prediction model; the weighting values of the C probabilistic predictive model are not adjusted.

S150, if customer information to be predicted input by the user terminal is received, obtaining customer variable information to be predicted corresponding to the customer information to be predicted based on the information conversion model.

And if customer information to be predicted input by the user terminal is received, obtaining customer variable information to be predicted corresponding to the customer information to be predicted based on the information conversion model. The customer information to be predicted can include one or more customer information, the customer information also includes account number, age, gender, occupation, total amount of assets, amount of purchased products, transaction times, transaction amount, latest transaction time and other information, the customer information to be predicted can include one or more customer information corresponding to the customer to be predicted, and each customer information in the customer information to be predicted can be converted through a customer rating rule and a variable conversion rule to obtain corresponding customer variable information to be predicted. The information of a certain client in the information of the clients to be predicted can be information corresponding to a newly added client or information corresponding to an existing client, and because the probability value of the client changes along with time, the probability values corresponding to a plurality of time points of the same client can be obtained to obtain the probability value curve.

In one embodiment, as shown in FIG. 5, step S150 includes sub-steps S151 and S152.

And S151, grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain the grade of the customer to be predicted.

And grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain the grade of the customer to be predicted. The customer rating rule is the rule information for obtaining the customer grade corresponding to each customer, and the specific rating manner is the same as the method in step S110, and the customer grade corresponding to each customer is obtained, so that the customer grade to be predicted corresponding to the customer information to be predicted can be obtained.

S152, converting each piece of customer information in the customer information to be predicted and the grade of the customer to be predicted according to the variable conversion rule and the conversion item to obtain the customer variable information to be predicted.

And converting each piece of customer information in the customer information to be predicted and the grade of the customer to be predicted according to the variable conversion rule and the conversion item to obtain the customer variable information to be predicted. In order to quantize each piece of client information in the piece of client information to be predicted and the client level corresponding to each piece of client information, each piece of client information needs to be converted into the piece of client variable information to be predicted through a variable conversion rule, namely, each piece of client information needs to be normalized, and the information needing to be converted comprises all pieces of client information contained in the piece of client information to be predicted and the client level corresponding to each piece of client information. Specifically, the variable conversion rule includes a rule for converting each item value included in the client information, and each item value corresponding to the conversion item can be converted to obtain a variable value, where the range of each variable value is [0, 1 ].

And S160, inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value.

And inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value. Specifically, the customer variable information to be predicted is respectively input into a plurality of probability prediction models contained in the combined probability prediction model, the prediction value of each probability prediction model is obtained through calculation in sequence, each probability prediction model is multiplied by the corresponding weight value, and the obtained products are accumulated to obtain the probability value of the customer information to be predicted. The probability value corresponding to the customer information to be predicted is a predicted value obtained by predicting whether the customer to be predicted has the same behavior as the customer of the positive sample or not based on the current consumption habits of the customers.

S170, the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information in the client information to be predicted is obtained and used as the target client.

And acquiring the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as a target client. And judging the probability value corresponding to each client to be predicted by taking the positive sample proportion value in the historical client information as a judgment threshold, if the probability value corresponding to a certain client to be predicted in the client information to be predicted is greater than the preset probability preset, indicating that the client has an obvious trend and is about to perform the same action as the positive sample client, and taking the client to be predicted as a target client. In addition, a judgment threshold for judging the probability value of the client to be predicted can be preset, namely, other fixed values are selected to replace a positive sample proportion value in the historical client information to be used as a judgment basis for judging the probability value.

In one embodiment, as shown in fig. 6, step S170 further includes sub-steps S171 and S172.

And S171, calculating a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information.

And calculating a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information. And after judging each piece of historical client information in the historical client information table based on the steps, obtaining sample information corresponding to each piece of historical client information, wherein the sample information is a positive sample or a negative sample, and calculating to obtain a positive sample proportion value corresponding to the client information base according to the sample information.

For example, if the number of the historical customer information corresponding to the sample information in the customer information base being a positive sample is 55, and the total number of the historical customer information included in the customer information base is 100, the corresponding positive sample ratio value 55/100 is 0.55.

And S172, judging the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value, so as to obtain the client to be predicted with the probability value larger than the positive sample proportion value as a target client.

And judging the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value so as to obtain the client to be predicted with the probability value larger than the positive sample proportion value as a target client. After the target client is obtained, further measures can be taken, for example, if the target client is a client with a loss tendency, corresponding prompt information can be pushed to the client to save the client.

In other embodiments of the present invention, step S180 is further included after step S170.

And S180, calculating a contrast score value of each customer to be predicted in the customer variable information to be predicted according to the variable information.

And calculating the contrast score value of each customer to be predicted in the customer variable information to be predicted according to the variable information. The comparison score value is a score value for comparing each client to be predicted in the variable information of the clients to be predicted based on the conversion item in the variable information, the comparison score value comprises a plurality of dimensions, and each dimension corresponds to one conversion item in the variable information. Through calculating the contrast score value corresponding to the client in the client variable information to be predicted and displaying the contrast score value in the radar map form, the contrast information of the client and the existing negative sample client on multiple dimensions can be known more intuitively.

In an embodiment, step S180 includes sub-steps S181, S182, and S183.

And S181, obtaining all the clients of which the sample information is the negative sample in the variable information to obtain the negative sample client variable.

And obtaining all the clients of which the sample information is the negative sample in the variable information to obtain the negative sample client variable. Based on the sample information in the customer variable information, the information corresponding to the customer with all the sample information being negative samples can be obtained and used as the negative sample customer variable.

And S182, calculating the average value of the conversion items corresponding to each conversion item in all the negative sample customer variables.

And calculating the average value of the conversion items corresponding to each conversion item in all the negative sample client variables. And calculating a comparison score value of the customer variable information to be predicted by taking the average value of the conversion items corresponding to each conversion item in the negative sample customer variable as a reference.

And S183, calculating the ratio of each customer in the customer variable information to be predicted to the average value of each conversion item to obtain the contrast score value of each customer in the customer variable information to be predicted.

And calculating the ratio of each customer in the customer variable information to be predicted to the average value of each conversion item to obtain the contrast score value of each customer in the customer variable information to be predicted. And calculating the ratio of the variable value of each client in the client variable information to be predicted to the average value of each conversion item to obtain a comparison score value containing a plurality of conversion items, wherein the comparison score value contains a plurality of dimensions, and each dimension corresponds to one conversion item in the variable information.

In the prediction method based on the client information provided by the embodiment of the invention, historical client information in a client information base is converted to obtain corresponding variable information, a plurality of probability prediction models corresponding to each preset prediction template are constructed according to the variable information, the plurality of probability prediction models are combined according to a combination formula to obtain a combined probability prediction model, the client information to be predicted is converted to obtain the client variable information to be predicted, the client variable information to be predicted is input into the trained combined probability prediction model to obtain a corresponding probability value, and a client with the probability value larger than a preset probability threshold value in the client information to be predicted is taken as a target client. By the method, the combined probability prediction model obtained based on the combination of the probability prediction models can greatly improve the efficiency and accuracy of obtaining the target customer based on customer information prediction.

The embodiment of the invention also provides a client information-based prediction device, which is used for executing any embodiment of the client information-based prediction method. Specifically, referring to fig. 7, fig. 7 is a schematic block diagram of a prediction apparatus based on customer information according to an embodiment of the present invention. The client information-based prediction apparatus may be configured in a user terminal.

As shown in fig. 7, the client information-based prediction apparatus 100 includes a client variable information acquisition unit 110, a combined probability prediction model construction unit 120, a probability prediction model combining unit 130, a model training unit 140, a client variable information acquisition unit 150 to be predicted, a probability value acquisition unit 160, and a target client acquisition unit 170.

A client variable information obtaining unit 110, configured to convert, according to a preset information conversion model, historical client information in a preset client information base to obtain variable information corresponding to the historical client information.

In another embodiment of the present invention, as shown in fig. 8, the client variable information obtaining unit 110 includes sub-units: a customer sample information acquisition unit 111, a customer level acquisition unit 112, and a variable conversion unit 113.

The client sample information obtaining unit 111 is configured to determine whether each piece of historical client information is a positive sample according to the sample determination rule, so as to obtain sample information corresponding to each piece of historical client information.

A customer grade obtaining unit 112, configured to obtain, according to the customer rating rule, a customer grade corresponding to each piece of historical customer information in the customer information base.

And a variable conversion unit 113, configured to convert, according to the variable conversion rule, an item value corresponding to a conversion item in the variable conversion rule, the customer sample information, and the customer level in a customer information base to obtain variable information corresponding to each piece of historical customer information.

And the combined probability prediction model constructing unit 120 is configured to construct a plurality of probability prediction models corresponding to each preset prediction template according to the variable information.

In other embodiments of the present invention, as shown in fig. 9, the combined probabilistic predictive model building unit 120 includes sub-units: an input node construction unit 121 and a probabilistic predictive model construction unit 122.

And the input node constructing unit 121 is configured to construct an input node according to the conversion item in the variable information.

And a probabilistic prediction model constructing unit 122, configured to construct a plurality of probabilistic prediction models according to each of the prediction templates and the input node.

And a probability prediction model combining unit 130, configured to combine a plurality of probability prediction models according to a preset combining formula to obtain a combined probability prediction model.

And the model training unit 140 trains the combined probability prediction model according to the variable information and preset model training rules to obtain the trained combined probability prediction model.

In other embodiments of the present invention, as shown in fig. 10, the model training unit 140 includes sub-units: a client variable information splitting unit 141, a probabilistic predictive model training unit 142, a parameter value determining unit 143, and a combined probabilistic predictive model training unit 144.

The client variable information splitting unit 141 is configured to split the variable information into data sets of corresponding quantities according to the data splitting information.

And a probabilistic prediction model training unit 142, configured to perform multiple rounds of training on all the probabilistic prediction models according to the parameter value adjustment rule and the multiple data sets to obtain a training result of each probabilistic prediction model in the combined probabilistic prediction model, where the training result includes an accuracy and a coverage rate corresponding to each probabilistic prediction model in each round of training.

And a parameter value determining unit 143, configured to select, according to the training result, a parameter value corresponding to a round of training in which the sum of the accuracy and the coverage of the probability prediction model is the highest, as a parameter value in the probability prediction model.

And a combined probability prediction model training unit 144, configured to train the combined probability prediction model according to the data set to obtain a trained combined probability prediction model.

And a to-be-predicted client variable information obtaining unit 150, configured to, if receiving the to-be-predicted client information input by the user terminal, obtain, based on the information conversion model, to-be-predicted client variable information corresponding to the to-be-predicted client information.

In another embodiment of the present invention, as shown in fig. 11, the to-be-predicted customer variable information obtaining unit 150 includes sub-units: a to-be-predicted customer rating unit 151 and a to-be-predicted customer information conversion unit 152.

And the to-be-predicted customer rating unit 151 is configured to rate each to-be-predicted customer in the to-be-predicted customer information according to a customer rating rule in the information conversion model to obtain a to-be-predicted customer level.

And grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain the grade of the customer to be predicted. The customer rating rule is rule information for obtaining the customer rating corresponding to each customer, and the specific rating mode is the same as the method adopted in the customer variable information obtaining unit 110, so that the customer rating corresponding to each customer can be obtained to obtain the customer rating corresponding to the customer information to be predicted

A to-be-predicted customer information conversion unit 152, configured to convert, according to the variable conversion rule and the conversion item, each piece of customer information in the to-be-predicted customer information and the to-be-predicted customer level to obtain to-be-predicted customer variable information.

And the probability value obtaining unit 160 is configured to input the information of the client variable to be predicted into the trained combined probability prediction model to obtain a corresponding probability value.

And the target client obtaining unit 170 is configured to obtain, as the target client, the client to be predicted, in the client information to be predicted, where the probability value is greater than the positive sample proportion value in the historical client information.

In another embodiment of the present invention, as shown in fig. 12, the target client obtaining unit 170 includes sub-units: a positive sample ratio value calculation unit 171 and a probability value judgment unit 172.

The positive sample proportion value calculating unit 171 is configured to calculate a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information.

And the probability value judging unit 172 is configured to judge the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value, so as to obtain the client to be predicted, of which the probability value is greater than the positive sample proportion value, as a target client.

In another embodiment of the present invention, the prediction apparatus 100 based on customer information further includes a sub-unit: the contrast score value calculation unit 180.

And the comparison score value calculation unit 180 is configured to calculate a comparison score value of each customer to be predicted in the customer variable information to be predicted according to the variable information.

In other embodiments of the present invention, the contrast score value calculating unit 180 includes sub-units: a negative sample customer variable acquisition unit 181, a conversion item average acquisition unit 182, and a calculation unit 183.

And a negative sample client variable obtaining unit 181, configured to obtain all clients whose sample information is a negative sample in the variable information to obtain a negative sample client variable.

A conversion item average value obtaining unit 182, configured to calculate a conversion item average value corresponding to each conversion item in all the negative sample customer variables.

The calculating unit 183 is configured to calculate a ratio between each customer in the customer variable information to be predicted and each average conversion item to obtain a comparison score value of each customer in the customer variable information to be predicted.

The prediction device based on the client information provided by the embodiment of the invention is used for executing the prediction method based on the client information, converting historical client information in a client information base to obtain corresponding variable information, constructing and obtaining a plurality of probability prediction models corresponding to each preset prediction template according to the variable information, combining the plurality of probability prediction models according to a combination formula to obtain a combined probability prediction model, converting the client information to be predicted to obtain the client variable information to be predicted, inputting the client variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value, and taking a client with the probability value larger than a preset probability threshold value in the client information to be predicted as a target client. By the method, the combined probability prediction model obtained based on the combination of the probability prediction models can greatly improve the efficiency and accuracy of obtaining the target customer based on customer information prediction.

The above-described client information-based prediction means may be implemented in the form of a computer program that can be run on a computer device as shown in fig. 13.

Referring to fig. 13, fig. 13 is a schematic block diagram of a computer device according to an embodiment of the present invention.

Referring to fig. 13, the computer device 500 includes a processor 502, memory, and a network interface 505 connected by a system bus 501, where the memory may include a non-volatile storage medium 503 and an internal memory 504.

The non-volatile storage medium 503 may store an operating system 5031 and a computer program 5032. The computer program 5032, when executed, may cause the processor 502 to perform a client information based prediction method.

The processor 502 is used to provide computing and control capabilities that support the operation of the overall computer device 500.

The internal memory 504 provides an environment for the execution of the computer program 5032 in the non-volatile storage medium 503, and when the computer program 5032 is executed by the processor 502, the processor 502 may be caused to execute a prediction method based on the client information.

The network interface 505 is used for network communication, such as providing transmission of data information. Those skilled in the art will appreciate that the configuration shown in fig. 13 is a block diagram of only a portion of the configuration associated with aspects of the present invention and is not intended to limit the computing device 500 to which aspects of the present invention may be applied, and that a particular computing device 500 may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

Wherein the processor 502 is configured to run the computer program 5032 stored in the memory to implement the following functions: converting historical customer information in a preset customer information base according to a preset information conversion model to obtain variable information corresponding to the historical customer information; constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information; combining a plurality of probability prediction models according to a preset combination formula to obtain a combined probability prediction model; training the combined probability prediction model according to the variable information and a preset model training rule to obtain a trained combined probability prediction model; if customer information to be predicted input by a user terminal is received, obtaining customer variable information to be predicted corresponding to the customer information to be predicted based on the information conversion model; inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value; and acquiring the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as a target client.

In one embodiment, the processor 502 performs the following operations when executing the step of converting the historical client information in the preset client information base according to the preset information conversion model to obtain the variable information corresponding to the historical client information: judging whether each piece of historical customer information is a positive sample according to the sample judgment rule so as to obtain sample information corresponding to each piece of historical customer information; obtaining a customer grade corresponding to each historical customer information in the customer information base according to the customer rating rule; and converting the item value corresponding to the conversion item in the variable conversion rule, the customer sample information and the customer grade in a customer information base according to the variable conversion rule to obtain variable information corresponding to each historical customer information.

In an embodiment, the processor 502, when executing the step of constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information, performs the following operations: constructing and obtaining an input node according to a conversion project in the customer variable information; and constructing a plurality of probability prediction models according to each prediction template and the input nodes.

In an embodiment, when the processor 502 performs the step of training the combined probabilistic predictive model according to the variable information and the preset model training rule to obtain the trained combined probabilistic predictive model, the following operations are performed: splitting the variable information into data sets with corresponding quantity according to the data splitting information; performing multiple rounds of training on all the probability prediction models according to the parameter value adjustment rule and the multiple data sets to obtain a training result of each probability prediction model in the combined probability prediction model, wherein the training result comprises the accuracy and the coverage rate corresponding to each probability prediction model in each round of training; selecting a parameter value corresponding to the highest round of training with the highest sum of the accuracy and the coverage rate corresponding to the probability prediction model as a parameter value in the probability prediction model according to the training result; and training the combined probability prediction model according to the data set to obtain the trained combined probability prediction model.

In an embodiment, when the processor 502 performs the step of acquiring the to-be-predicted client variable information corresponding to the to-be-predicted client information based on the information conversion model if the to-be-predicted client information input by the user terminal is received, the following operations are performed: grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain a grade of the customer to be predicted; and converting the information of each customer in the customer information to be predicted and the grade of the customer to be predicted according to the variable conversion rule and the conversion item to obtain the variable information of the customer to be predicted.

In an embodiment, the processor 502 performs the following operations when performing the step of acquiring the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as the target client: calculating a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information; and judging the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value so as to obtain the client to be predicted with the probability value larger than the positive sample proportion value as a target client.

Those skilled in the art will appreciate that the embodiment of a computer device illustrated in fig. 13 does not constitute a limitation on the specific construction of the computer device, and that in other embodiments a computer device may include more or fewer components than those illustrated, or some components may be combined, or a different arrangement of components. For example, in some embodiments, the computer device may only include a memory and a processor, and in such embodiments, the structures and functions of the memory and the processor are consistent with those of the embodiment shown in fig. 13, and are not described herein again.

It should be understood that, in the embodiment of the present invention, the Processor 502 may be a Central Processing Unit (CPU), and the Processor 502 may also be other general purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable gate arrays (FPGAs) or other Programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. Wherein a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

In another embodiment of the invention, a computer-readable storage medium is provided. The computer readable storage medium may be a non-volatile computer readable storage medium. The computer-readable storage medium stores a computer program, wherein the computer program when executed by a processor implements the steps of: converting historical customer information in a preset customer information base according to a preset information conversion model to obtain variable information corresponding to the historical customer information; constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information; combining a plurality of probability prediction models according to a preset combination formula to obtain a combined probability prediction model; training the combined probability prediction model according to the variable information and a preset model training rule to obtain a trained combined probability prediction model; if customer information to be predicted input by a user terminal is received, obtaining customer variable information to be predicted corresponding to the customer information to be predicted based on the information conversion model; inputting the customer variable information to be predicted into the trained combined probability prediction model to obtain a corresponding probability value; and acquiring the client to be predicted with the probability value larger than the positive sample proportion value in the historical client information as a target client.

In an embodiment, the step of converting the historical customer information in the preset customer information base according to a preset information conversion model to obtain variable information corresponding to the historical customer information includes: judging whether each piece of historical customer information is a positive sample according to the sample judgment rule so as to obtain sample information corresponding to each piece of historical customer information; obtaining a customer grade corresponding to each historical customer information in the customer information base according to the customer rating rule; and converting the item value corresponding to the conversion item in the variable conversion rule, the customer sample information and the customer grade in a customer information base according to the variable conversion rule to obtain variable information corresponding to each historical customer information.

In an embodiment, the step of constructing a plurality of probability prediction models corresponding to each preset prediction template according to the variable information includes: constructing and obtaining an input node according to a conversion project in the customer variable information; and constructing a plurality of probability prediction models according to each prediction template and the input nodes.

In an embodiment, the step of training the combined probability prediction model according to the variable information and a preset model training rule to obtain a trained combined probability prediction model includes: splitting the variable information into data sets with corresponding quantity according to the data splitting information; splitting the variable information into data sets with corresponding quantity according to the data splitting information; performing multiple rounds of training on all the probability prediction models according to the parameter value adjustment rule and the multiple data sets to obtain a training result of each probability prediction model in the combined probability prediction model, wherein the training result comprises the accuracy and the coverage rate corresponding to each probability prediction model in each round of training; selecting a parameter value corresponding to the highest round of training with the highest sum of the accuracy and the coverage rate corresponding to the probability prediction model as a parameter value in the probability prediction model according to the training result; and training the combined probability prediction model according to the data set to obtain the trained combined probability prediction model.

In an embodiment, the step of obtaining, based on the information conversion model, to-be-predicted client variable information corresponding to the to-be-predicted client information if the to-be-predicted client information input by the user terminal is received includes: grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain a grade of the customer to be predicted; and converting the information of each customer in the customer information to be predicted and the grade of the customer to be predicted according to the variable conversion rule and the conversion item to obtain the variable information of the customer to be predicted.

In an embodiment, the step of acquiring the to-be-predicted client with the probability value greater than the positive sample proportion value in the historical client information as the target client includes: calculating a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information; and judging the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value so as to obtain the client to be predicted with the probability value larger than the positive sample proportion value as a target client.

It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described apparatuses, devices and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again. Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the examples described in connection with the embodiments disclosed herein may be embodied in electronic hardware, computer software, or combinations of both, and that the components and steps of the examples have been described in a functional general in the foregoing description for the purpose of illustrating clearly the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. 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 by the present invention, it should be understood that the disclosed apparatus, device and method can be implemented in other ways. For example, the above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only a logical division, and there may be other divisions when the actual implementation is performed, or units having the same function may be grouped into one unit, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may also be an electric, mechanical or other form of connection.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention essentially contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product stored in a computer-readable storage medium, which includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned computer-readable storage media comprise: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a magnetic disk, or an optical disk.

While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims

1. A prediction method based on customer information, comprising:

2. The customer information-based prediction method according to claim 1, wherein the information transformation model includes a customer rating rule, a sample judgment rule and a variable transformation rule, and the transforming the historical customer information in the preset customer information base according to the preset information transformation model to obtain the variable information corresponding to the historical customer information comprises:

judging whether each piece of historical customer information is a positive sample according to the sample judgment rule so as to obtain sample information corresponding to each piece of historical customer information;

obtaining a customer grade corresponding to each historical customer information in the customer information base according to the customer rating rule;

and converting the item value corresponding to the conversion item in the variable conversion rule, the customer sample information and the customer grade in a customer information base according to the variable conversion rule to obtain variable information corresponding to each historical customer information.

3. The customer information-based prediction method according to claim 1, wherein the constructing a plurality of probabilistic prediction models corresponding to each preset prediction template according to the variable information comprises:

constructing and obtaining an input node according to a conversion project in the customer variable information;

and constructing a plurality of probability prediction models according to each prediction template and the input nodes.

4. The customer information-based prediction method of claim 1, wherein the model training rules include data splitting information and parameter value adjustment rules, and the training of the combined probabilistic predictive model according to the variable information and preset model training rules to obtain the trained combined probabilistic predictive model comprises:

splitting the variable information into data sets with corresponding quantity according to the data splitting information;

performing multiple rounds of training on all the probability prediction models according to the parameter value adjustment rule and the multiple data sets to obtain a training result of each probability prediction model in the combined probability prediction model, wherein the training result comprises the accuracy and the coverage rate corresponding to each probability prediction model in each round of training;

selecting a parameter value corresponding to the highest round of training with the highest sum of the accuracy and the coverage rate corresponding to the probability prediction model as a parameter value in the probability prediction model according to the training result;

and training the combined probability prediction model according to the data set to obtain the trained combined probability prediction model.

5. The client information-based prediction method according to claim 1, wherein the obtaining of the client variable information to be predicted corresponding to the client information to be predicted based on the information conversion model includes:

grading each customer to be predicted in the customer information to be predicted according to a customer grading rule in the information conversion model to obtain a grade of the customer to be predicted;

and converting the information of each client in the client information to be predicted and the grade of the client to be predicted according to a variable conversion rule in the information conversion model and a conversion item in the variable conversion rule to obtain the variable information of the client to be predicted.

6. The client information-based prediction method according to claim 2, wherein the obtaining, as the target client, the client to be predicted having the probability value greater than the positive sample proportion value in the historical client information, comprises:

calculating a positive sample proportion value corresponding to the customer information base according to the sample information of each piece of historical customer information;

and judging the probability value of the client to be predicted in the client information to be predicted according to the positive sample proportion value so as to obtain the client to be predicted with the probability value larger than the positive sample proportion value as a target client.

7. A prediction apparatus based on customer information, comprising:

8. The client information-based prediction apparatus according to claim 7, wherein the client variable information acquisition unit includes:

the client sample information acquisition unit is used for judging whether each piece of historical client information is a positive sample according to the sample judgment rule so as to acquire sample information corresponding to each piece of historical client information;

the client grade acquisition unit is used for acquiring a client grade corresponding to each piece of historical client information in the client information base according to the client rating rule;

and the variable conversion unit is used for converting the item values corresponding to the conversion items in the variable conversion rules, the client sample information and the client levels in the client information base according to the variable conversion rules to obtain the variable information corresponding to each historical client information.

9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the client information based prediction method according to any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program which, when executed by a processor, causes the processor to execute the client information-based prediction method according to any one of claims 1 to 6.