CN115203556A - Score prediction model training method and device, electronic equipment and storage medium - Google Patents
Score prediction model training method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN115203556A CN115203556A CN202210836399.1A CN202210836399A CN115203556A CN 115203556 A CN115203556 A CN 115203556A CN 202210836399 A CN202210836399 A CN 202210836399A CN 115203556 A CN115203556 A CN 115203556A
- Authority
- CN
- China
- Prior art keywords
- score
- user
- scores
- prediction model
- real
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000012549 training Methods 0.000 title claims abstract description 113
- 238000000034 method Methods 0.000 title claims abstract description 69
- 230000006399 behavior Effects 0.000 claims description 56
- 238000012545 processing Methods 0.000 claims description 37
- 230000015654 memory Effects 0.000 claims description 21
- 230000008447 perception Effects 0.000 abstract description 13
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 14
- 230000000694 effects Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000004590 computer program Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 235000019800 disodium phosphate Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Operations Research (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a scoring prediction model training method, a device, electronic equipment and a storage medium, which relate to the technical field of artificial intelligence and are used for solving the problem that the prediction result of a scoring prediction model trained at the present stage is not accurate enough, and the method comprises the following steps: acquiring sample data, wherein the sample data comprises multidimensional behavior characteristics of each user in a plurality of users and real user scores corresponding to the multidimensional behavior characteristics of each user; determining the occurrence times of each real user score in the sample data; determining a target user score corresponding to the multi-dimensional behavior characteristics of each user according to the frequency of occurrence of each real user score in sample data; and training the initial score prediction model according to the target user score corresponding to the multi-dimensional behavior characteristics of each user and the multi-dimensional behavior characteristics of each user to obtain a trained target score prediction model. The method and the device are used for scoring prediction of user network perception.
Description
Technical Field
The application relates to the technical field of artificial intelligence, in particular to a scoring prediction model training method and device, electronic equipment and a storage medium.
Background
Improving mobile network awareness scores is a continuing concern for operators. The operator generally sets the score value to 10 numbers of 1 to 10, so that the user can score according to the satisfaction degree of the user. At present, operators generally adopt a score prediction model for network perception of users in the whole network in combination with an intelligent algorithm to accurately predict the network perception scores of the users aiming at the current situation that the network perception scores of the users are low and the passive network problem solving cannot achieve a good effect. Furthermore, for the users with lower prediction scores, the operators actively adopt a key guarantee method to achieve the purpose of improving the perception of the users in the whole network. However, in the training method of the score prediction model at the present stage, the prediction result of the trained score prediction model is not accurate enough.
Disclosure of Invention
The application provides a score prediction model training method and device, an electronic device and a storage medium, which can solve the problem that the prediction result of a score prediction model trained at the present stage is not accurate enough.
In order to achieve the purpose, the following technical scheme is adopted in the application:
in a first aspect, the present application provides a score prediction model training method, including: acquiring sample data, wherein the sample data comprises multidimensional behavior characteristics of each user in a plurality of users and real user scores corresponding to the multidimensional behavior characteristics of each user; determining the number of times each real user score appears in the sample data; determining a target user score corresponding to the multi-dimensional behavior characteristics of each user according to the frequency of occurrence of each real user score in sample data; the target user score is used for representing a predicted value of the score of each user; and training the initial score prediction model according to the target user score corresponding to the multi-dimensional behavior characteristics of each user and the multi-dimensional behavior characteristics of each user to obtain a trained target score prediction model.
Based on the technical scheme, the initial prediction score determined by the initial score prediction model is corrected according to the frequency of occurrence of each user score in sample data, the final target user score is determined, and the initial score prediction model is trained according to the final target user score to obtain the trained target score prediction model. Therefore, the trained target scoring prediction model is higher in accuracy and smaller in MAE value.
In a possible implementation manner, determining a target user score corresponding to the multidimensional behavior feature of each user according to the number of times that each real user score appears in sample data specifically includes: determining the distribution proportion of each real user score according to the occurrence frequency of each real user score in the sample data; rounding the initial prediction scores according to the distribution proportion of each real user score, and determining a target user score corresponding to the multi-dimensional behavior characteristics of each user; and the initial prediction score is the user score determined by the initial score prediction model.
In a possible implementation manner, rounding the initial prediction score according to the distribution ratio of each real user score includes: determining the distribution proportion of scores of every two real users with adjacent sizes; according to the distribution proportion of each two real user scores adjacent in size, carrying out local rounding on the initial prediction scores; according to the distribution proportion of each two real user scores adjacent in size, the initial prediction score is locally rounded to meet the following formula:
wherein A represents a target user score, B 1 And B 2 Representing each two real user scores of adjacent size, N representing the initial prediction score, x 1 And x 2 Indicating the distribution ratio of every two real user scores with adjacent sizes.
In a possible implementation manner, after the initial prediction score is locally rounded according to a distribution ratio of each two real user scores with adjacent sizes, the method further includes: determining the distribution proportion of any two real user scores; according to the distribution ratio of any two real user scores, performing global rounding on the initial prediction score after the local rounding; according to the distribution proportion of any two real users, the following formula is satisfied by carrying out global rounding on the initial prediction scores after the local rounding is carried out:
wherein A represents a target user score, B 3 And B 4 Representing any two true user scores, N representing the initial predicted score, x 3 And x 4 Representing the distribution ratio of any two real user scores.
In a possible implementation manner, the multidimensional behavior characteristic of each user is determined based on the gender and age of each user, the number of times of using networks of different network systems, the length of time of using networks of different network systems, the network call completing rate, and the average delay of the downlink round trip time RTT.
In a second aspect, the present application provides a score prediction model training apparatus, including: the score prediction model training device comprises: an acquisition unit and a processing unit; the system comprises an acquisition unit, a display unit and a display unit, wherein the acquisition unit is used for acquiring sample data, and the sample data comprises the multidimensional behavior characteristics of each user in a plurality of users and real user scores corresponding to the multidimensional behavior characteristics of each user; the processing unit is used for determining the frequency of occurrence of each real user score in the sample data; the processing unit is also used for determining a target user score corresponding to the multi-dimensional behavior characteristics of each user according to the frequency of occurrence of each real user score in the sample data; the target user score is used for representing a predicted value of the score of each user; and the processing unit is also used for training the initial scoring prediction model according to the target user score corresponding to the multi-dimensional behavior characteristic of each user and the multi-dimensional behavior characteristic of each user to obtain a trained target scoring prediction model.
In a possible implementation manner, the processing unit is further configured to determine a distribution ratio of each real user score according to the number of times that each real user score appears in the sample data; the processing unit is also used for rounding the initial prediction scores according to the distribution proportion of each real user score and determining a target user score corresponding to the multi-dimensional behavior characteristics of each user; and the initial prediction score is the user score determined by the initial score prediction model.
In a possible implementation manner, the processing unit is further configured to determine a distribution ratio of scores of every two real users whose sizes are adjacent to each other; the processing unit is also used for carrying out local rounding on the initial prediction scores according to the distribution proportion of every two adjacent real user scores; according to the distribution proportion of each two real user scores adjacent in size, the initial prediction score is locally rounded to meet the following formula:
wherein A represents a target user score, B 1 And B 2 Representing each two real user scores adjacent in size, N representing initial pre-scoreMeasurement score, x 1 And x 2 Indicating the distribution ratio of every two real user scores with adjacent sizes.
In a possible implementation manner, the processing unit is further configured to determine a distribution ratio of any two real user scores; the processing unit is also used for carrying out global rounding on the initial prediction scores after the local rounding according to the distribution proportion of any two real user scores; according to the distribution proportion of any two real users, the following formula is satisfied by carrying out global rounding on the initial prediction scores after the local rounding is carried out:
wherein A represents a target user score, B 3 And B 4 Representing any two true user scores, N representing the initial predicted score, x 3 And x 4 Representing the distribution ratio of any two real user scores.
In a possible implementation manner, the multidimensional behavior characteristic of each user is determined based on the gender and age of each user, the number of times of using networks of different network standards, the duration of using networks of different network standards, the network call completing rate and the average delay of the downlink round trip time RTT.
In addition, the technical effect of the score prediction model training method according to the second aspect may refer to the technical effect of the score prediction model training method described in the first aspect, and is not repeated here.
In a third aspect, the present application provides a computer readable storage medium storing one or more programs, the one or more programs comprising instructions, which when executed by an electronic device of the present application, cause the electronic device to perform the score prediction model training method as described in the first aspect and any one of the possible implementations of the first aspect.
In a fourth aspect, the present application provides an electronic device comprising: a processor and a memory; wherein the memory is used for storing one or more programs, the one or more programs comprising computer executable instructions, which when executed by the processor, cause the electronic device to perform the scoring prediction model training method as described in the first aspect and any possible implementation manner of the first aspect.
In a fifth aspect, the present application provides a computer program product comprising instructions that, when executed on a computer, cause an electronic device of the present application to perform a scoring predictive model training method as described in the first aspect and any one of the possible implementations of the first aspect.
In a sixth aspect, the present application provides a chip system, which is applied to a scoring prediction model training device; the system-on-chip includes one or more interface circuits, and one or more processors. The interface circuit and the processor are interconnected through a line; the interface circuit is configured to receive a signal from a memory of the scoring predictive model training device and send the signal to the processor, the signal including computer instructions stored in the memory. When the processor executes the computer instructions, the scoring prediction model training device executes the scoring prediction model training method according to the first aspect and any possible design thereof.
In the present application, the names of the above-mentioned score prediction model training devices do not limit the devices or functional units themselves, and in practical implementations, these devices or functional units may appear by other names. Insofar as the functions of the respective devices or functional units are similar to those of the present application, they are within the scope of the claims of the present application and their equivalents.
Drawings
Fig. 1 is a schematic structural diagram of a scoring prediction model training device according to an embodiment of the present disclosure;
fig. 2 is a schematic flowchart of a score prediction model training method according to an embodiment of the present disclosure;
fig. 3 is a schematic flowchart of another score prediction model training method according to an embodiment of the present disclosure;
fig. 4 is a schematic flowchart of another scoring prediction model training method provided in the embodiment of the present application;
fig. 5 is a schematic structural diagram of a scoring prediction model training device according to an embodiment of the present disclosure;
fig. 6 is a schematic structural diagram of another score prediction model training device according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. 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 application.
The character "/" herein generally indicates that the former and latter associated objects are in an "or" relationship. For example, A/B may be understood as A or B.
The terms "first" and "second" in the description and claims of the present application are used for distinguishing between different objects and not for describing a particular order of the objects. For example, the first edge service node and the second edge service node are used for distinguishing different edge service nodes, and are not used for describing the characteristic sequence of the edge service nodes.
Furthermore, the terms "including" and "having," and any variations thereof, as referred to in the description of the present application, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not limited to only those steps or elements but may alternatively include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In addition, in the embodiments of the present application, words such as "exemplarily" or "for example" are used for indicating as examples, illustrations or explanations. Any embodiment or design described herein as "exemplary" or "e.g.," is not necessarily to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the word "exemplary" or "e.g.," is intended to present concepts in a concrete fashion.
Currently, improving the mobile network awareness score is a problem that operators are continuously concerned about. The operator generally sets the score value to 10 numbers of 1 to 10, so that the user can score according to the satisfaction degree of the user. The user score is 10, which indicates that the user is very satisfied; the user score was 1, indicating that the user was very dissatisfied.
Aiming at the current situations that the user network perception score is low and the passive solution to the network problem cannot achieve a good effect, operators generally adopt a combination intelligent algorithm to establish a score prediction model of the whole network user network perception so as to accurately predict the user network perception score. Further, for users with low prediction scores, operators actively adopt a key guarantee method to achieve the purpose of improving the perception of users in the whole network.
To measure the accuracy of the prediction model of the network perceptual score, the evaluation is generally performed by an average Absolute Error MAE (MAE) value, which describes an average of Absolute errors. For the scoring prediction model related to the application, the smaller the MAE value is, namely, the closer the scoring result representing model prediction is to the actual scoring result of the user, the better the prediction effect is.
At the present stage, when a network perception score prediction model is established, two ideas are generally adopted: classification and regression. When the grading prediction of user network perception is carried out through a classification method, the target value is 1-10 and 10 grading values in total. The user characteristics are used as model input, and the score of the user can be predicted to belong to a certain category of 10 numbers by training a classification model. However, classification also has some disadvantages: according to the analysis of actual data, the situation that the distribution of the scores of the users is extremely unbalanced is found, the number of the scores of 10 and 1 is the largest, and the number of the intermediate scores such as 3 and 4 is small, so that the situation of misclassification is easily caused during model training. The classification method is adopted to carry out the grading prediction of the user network perception, and once the classification is wrong, the evaluation index MAE value and the prediction accuracy are greatly influenced.
When the user network-aware score prediction is performed by regression, the target value is still 10 different scores. And then the user characteristics are used as input to train the regression model, so that the score of the user score can be predicted. However, regression methods also have some disadvantages: the regression method usually sets an evaluation index, namely an MAE value index, as an objective function of the model, and the value of the objective function tends to become smaller in the training process of the model; most of the predicted results of the models trained based on the principle are not integers, but decimal numbers such as 3.2,5.8 and the like, and the situation cannot occur when a user actually scores the scores, the deviation is generated from the actual situation, and the accuracy of the predicted results is poor.
Illustratively, in response to the above problem, the following two schemes are provided at present:
according to the scheme I, a user score prediction model is constructed through a deep learning technology; acquiring the purchase history and scoring records of the user on the tourism resources, and training a scoring prediction model of the user by using the tourism resources purchased by the user; and inputting the tourism resources which are not purchased by the user into a user score prediction model to obtain a prediction score, and recommending the first c tourism resources which have the highest prediction score and are not purchased by the user to the user. The method does not consider the existence of respective possible defects of classification and regression in a user scoring prediction model, and does not carry out optimization.
Obtaining a sample song, wherein the sample song comprises a pre-labeled song label; acquiring a user scoring matrix, wherein the user scoring matrix comprises scores of at least one user for the sample songs, and the scores are calculated according to the operation behaviors of the user for the sample songs; generating a song classifier according to the user scoring matrix and the song label of the sample song; song tags are assigned to the individual songs in the song library by the song classifier. But does not take into account the influence on the accuracy of song classification due to unbalanced scoring by the user.
The scheme can not solve the problem that the user scoring prediction accuracy is low when the scoring prediction of user network perception is carried out by a regression method or a classification method. In general, in the training method of the score prediction model at the present stage, the prediction result of the trained score prediction model is not accurate enough.
In order to solve the problem that the prediction result of the trained score prediction model is not accurate enough in the training method of the score prediction model at the present stage, the method for training the score prediction model is provided, and the user behavior characteristics are used as input data to predict the score of the user based on the classic machine learning regression model Lightgbm. And optimally rounding and reclassifying the predicted decimal result to form a user scoring prediction model with higher accuracy and smaller MAE value.
Exemplarily, as shown in fig. 1, a schematic structural diagram of a scoring prediction model training device provided in the present application is shown. The score prediction model training device 10 specifically includes: the system comprises a sample data module 11, a user grading processing module 12 and a model training module 13.
The sample data module 11 is configured to obtain sample data. The sample data comprises the multi-dimensional behavior characteristics of each user in the multiple users and the real user scores corresponding to the multi-dimensional behavior characteristics of each user. The sample data module 11 is further configured to determine the number of times each real user score appears in the sample data.
And the user score processing module 12 is configured to determine, according to the number of times that each real user score appears in the sample data, a target user score corresponding to the multidimensional behavior feature of each user.
And the model training module 13 is configured to construct an initial score prediction model, and train the initial score prediction model according to the target user score corresponding to the multidimensional behavior feature of each user and the multidimensional behavior feature of each user, so as to obtain a trained target score prediction model.
It should be noted that, in the scoring prediction model training method provided in the present application, the executing subject is a scoring prediction model training device. The scoring prediction model training device can be an electronic device (such as a computer terminal and a server), a processor in the electronic device, a control module used for scoring prediction model training in the electronic device, and a client used for scoring prediction model training in the electronic device.
The following describes a flow of the score prediction model training method provided in this embodiment.
Illustratively, as shown in fig. 2, a scoring prediction model training method provided for the present application specifically includes the following steps S201 to S204:
s201, the score prediction model training device obtains sample data.
The sample data comprises the multi-dimensional behavior characteristics of each user in the multiple users and the real user scores corresponding to the multi-dimensional behavior characteristics of each user.
Optionally, the multidimensional behavior characteristic of each user is determined according to the gender and age of each user, the number of times of using networks of different network systems, the use duration of the networks of different network systems, the network call completing rate and the average downlink Round-Trip Time (RTT).
Illustratively, the score prediction model training device can acquire sample data through the operator data bureau.
Optionally, the real user score is set to 10 numbers from 1 to 10, and is obtained by scoring according to the satisfaction degree of the user. The user score is 10, which indicates that the user is very satisfied; the user score was 1, indicating that the user was very dissatisfied.
It is understood that the real user score is a real score value of each user to the network provided by the operator based on the multi-dimensional behavior characteristics of the user.
S202, the score prediction model training device determines the occurrence frequency of each real user score in sample data.
Optionally, after obtaining the sample data, the score prediction model training device counts the occurrence frequency of each score in the sample data according to the setting condition of the real user score. Illustratively, in connection with the example in S201, the score prediction model training device counts the number of occurrences in the sample data in 1 to 10 points.
S203, determining a target user score corresponding to the multi-dimensional behavior characteristics of each user by the score prediction model training device according to the occurrence frequency of each real user score in the sample data.
Optionally, the score prediction model training device determines the distribution ratio of each real user score according to the number of times that each real user score appears in the sample data.
Further, the score prediction model training device rounds the initial prediction scores according to the distribution proportion of each real user score, and determines the target user score corresponding to the multi-dimensional behavior feature of each user. It is understood that, in the present embodiment, the target user score is a predicted value for representing the score for each user.
It will be appreciated that the initial prediction scores described above are user scores determined by the initial score prediction model. The initial score prediction model is a model constructed based on a classical machine learning regression model Lightgbm, and the user score predicted according to the sample data is the initial prediction score. The initial prediction score is based on the characteristics of the algorithm itself, and the score value may have a decimal place. In the actual user scoring process, the scoring value does not have decimal place, so the scoring prediction model training device in the application can perform rounding processing on the initial prediction scoring.
In a possible implementation manner, the score prediction model training device performs local rounding on the initial prediction scores according to the distribution ratio of each two adjacent real user scores. The process of the specific score prediction model training device performing local rounding on the initial prediction score is described in S301-S302 below, and is not described herein again.
In another possible implementation manner, the score prediction model training device performs local rounding on the initial prediction scores according to the distribution ratio of every two real user scores adjacent in size, and then performs global rounding on the initial prediction scores after the local rounding according to the distribution ratio of any two real user scores. The process of the specific score prediction model training device performing local rounding and global rounding on the initial prediction score is referred to the following S401-S402, which is not described herein again.
It can be understood that after the score prediction model training device rounds the initial prediction score, the prediction score is more fit to the actual situation, and therefore the accuracy of score prediction is improved.
And S204, training the initial scoring prediction model by the scoring prediction model training device according to the scoring of the target user corresponding to the multi-dimensional behavior characteristics of each user and the multi-dimensional behavior characteristics of each user to obtain a trained target scoring prediction model.
Optionally, the score prediction model training device calculates the MAE value index according to a target user score corresponding to the multidimensional behavior feature of each user, that is, the target user score is used as a prediction result of the initial score prediction model, and determines whether the MAE value meets a preset requirement. For example, the preset requirement that the MAE value satisfies may be that the MAE value is less than or equal to a preset value, or that a function related to the MAE value converges, which is not specifically limited in this embodiment of the present application.
It can be understood that, if the MAE value index of the initial scoring prediction model meets the preset requirement after the scoring prediction model training device takes the score of the target user as the prediction result of the initial scoring prediction model, the scoring prediction model training device determines that the training of the initial scoring prediction model is completed, and determines the initial scoring prediction model at this time as the trained target scoring prediction model.
Similarly, if the MAE value index of the initial scoring prediction model does not meet the preset requirement after the scoring prediction model training device takes the scoring of the target user as the prediction result of the initial scoring prediction model, the iterative training of the initial scoring prediction model is continued according to the sample data. In each iterative training, the score prediction model training device rounds the prediction result of the initial score prediction model according to the occurrence frequency of each real user score in sample data, so as to determine the target user score corresponding to the multi-dimensional behavior characteristic of each user. And then, the score prediction model training device takes the score of the target user determined in the iteration as a prediction result of the initial score prediction model until the MAE value index of the initial score prediction model meets the preset requirement.
Based on the technical scheme, the initial prediction score determined by the initial score prediction model is corrected according to the frequency of each user score appearing in sample data, the final target user score is determined, and the initial score prediction model is trained according to the final target user score to obtain the trained target score prediction model. Therefore, the trained target score prediction model has higher accuracy and smaller MAE value.
Exemplarily, with reference to fig. 2 and as shown in fig. 3, in the score prediction model training method provided by the present application, the score prediction model training device determines, according to the number of times that each real user score appears in sample data, a target user score corresponding to the multidimensional behavior feature of each user, and specifically includes the following steps S301 to S302:
s301, the score prediction model training device determines the distribution ratio of scores of every two adjacent real users.
It can be understood that the score prediction model training device determines the distribution ratio of every two adjacent real user scores in size in order to round the decimal place existing in the initial prediction score. If the two real user scores determined by the score prediction model training device are not adjacent in size, the number of the initial prediction scores is also changed, which is not in accordance with the purpose of rounding the small numbers existing in the initial prediction scores in the embodiment.
It should be noted that, for a specific method for determining a distribution ratio of a certain real user score, refer to the foregoing S202, which is not described herein again.
S302, the score prediction model training device performs local rounding on the initial prediction scores according to the distribution proportion of each two adjacent real user scores.
In a possible implementation manner, the score prediction model training device performs local rounding on the initial prediction scores according to the distribution ratio of scores of every two real users adjacent in size, and the formula is as follows:
wherein A represents a target user score, B 1 And B 2 Representing each two real user scores that are adjacent in size, N representing the initial predicted score, x 1 And x 2 Indicating the distribution ratio of every two real user scores with adjacent sizes.
In the scoring prediction model training method provided by the present application, the scoring prediction model training device determines the specific process of the target user score corresponding to the multidimensional behavior feature of each user according to the number of times that each real user score appears in the sample data.
Exemplarily, with reference to fig. 3 and as shown in fig. 4, in the score prediction model training method provided by the present application, the score prediction model training device performs local rounding on the initial prediction score according to the distribution ratio of each two real user scores whose sizes are adjacent to each other, and then performs global rounding on the initial prediction score after the local rounding according to the distribution ratio of any two real user scores, which specifically includes the following steps S401 to S402:
s401, the score prediction model training device determines the distribution ratio of scores of any two real users.
It can be understood that the score prediction model training device determines the distribution ratio of any two real user scores, and aims to perform global rounding on the initial prediction score and achieve the purpose of performing re-splitting on the initial prediction score.
It should be noted that, for a specific method for determining a distribution ratio of a certain real user score, refer to the foregoing S202, which is not described herein again.
S402, the score prediction model training device performs global rounding on the initial prediction scores after the local rounding according to the distribution ratio of any two real user scores.
In a possible implementation manner, the score prediction model training device performs global rounding on the initial prediction score after performing the local rounding according to the distribution ratio of any two real users, and the following formula is satisfied:
wherein A represents the target user score, B 3 And B 4 Representing the arbitrary two real user scores, N representing the initial predicted score, x 3 And x 4 Representing a distribution ratio of the arbitrary two real user scores.
It should be understood that based on the foregoing S301-S302 and S401-S402, the score prediction model training device determines the target user score corresponding to the multidimensional behavior feature of each user, which is also equivalent to determining a clustering center, and the clustering center can embody which scores are specifically aggregated according to the actual user score obtained by sample data analysis, so as to perform proportional reclassification on the prediction result of the user predicted by the initial score prediction model, so as to reduce the MAE index of the prediction result of the score prediction model, and improve the accuracy of score prediction.
In the embodiment of the present application, the score prediction model training device may be divided into function modules or function units according to the above method examples, for example, each function module or function unit may be divided corresponding to each function, or two or more functions may be integrated into one processing module. The integrated module may be implemented in a form of hardware, or may be implemented in a form of a software functional module or a functional unit. The division of the modules or units in the embodiment of the present application is schematic, and is only a logic function division, and there may be another division manner in actual implementation.
Fig. 5 is a schematic diagram illustrating a possible structure of a scoring prediction model training apparatus according to an embodiment of the present application. The score prediction model training apparatus 500 includes: an acquisition unit 501 and a processing unit 502.
The obtaining unit 501 is configured to obtain sample data, where the sample data includes a multidimensional behavior feature of each user in multiple users and a real user score corresponding to the multidimensional behavior feature of each user.
A processing unit 502 for determining the number of times each real user score appears in the sample data.
The processing unit 502 is further configured to determine, according to the number of times that each real user score appears in the sample data, a target user score corresponding to the multidimensional behavior feature of each user; the target user score is used to characterize the predicted value of the score for each user.
The processing unit 502 is further configured to train the initial score prediction model according to the target user score corresponding to the multidimensional behavior feature of each user and the multidimensional behavior feature of each user, so as to obtain a trained target score prediction model.
Optionally, the processing unit 502 is further configured to determine a distribution ratio of each real user score according to the number of times that each real user score appears in the sample data.
Optionally, the processing unit 502 is further configured to round the initial prediction score according to a distribution ratio of each real user score, and determine a target user score corresponding to the multidimensional behavior feature of each user; and the initial prediction score is the user score determined by the initial score prediction model.
Optionally, the processing unit 502 is further configured to determine a distribution ratio of each two real user scores with adjacent sizes.
Optionally, the processing unit 502 is further configured to perform local rounding on the initial prediction scores according to a distribution ratio of each two real user scores with adjacent sizes; according to the distribution proportion of each two real user scores adjacent in size, the initial prediction score is locally rounded to meet the following formula:
wherein A represents a target user score, B 1 And B 2 Representing each two real user scores that are adjacent in size, N representing the initial predicted score, x 1 And x 2 Indicating the distribution ratio of every two real user scores with adjacent sizes.
Optionally, the processing unit 502 is further configured to determine a distribution ratio of any two real user scores.
Optionally, the processing unit 502 is further configured to perform global rounding on the initial prediction score after performing local rounding according to a distribution ratio of any two real user scores; according to the distribution proportion of any two real users, the following formula is satisfied by carrying out global rounding on the initial prediction scores after the local rounding is carried out:
wherein A represents a target user score, B 3 And B 4 Representing any two true user scores, N representing the initial predicted score, x 3 And x 4 Representing the distribution ratio of any two real user scores.
Optionally, the score prediction model training apparatus 500 may further include a storage unit (shown by a dashed box in fig. 5) that stores a program or instructions, and when the processing unit 501 executes the program or instructions, the score prediction model training apparatus may perform the score prediction model training method according to the above method embodiment.
In addition, for the technical effect of the score prediction model training apparatus described in fig. 5, reference may be made to the technical effect of the score prediction model training method described in the foregoing embodiment, and details are not repeated here.
Fig. 6 is a schematic diagram of another possible structure of the score prediction model training apparatus according to the foregoing embodiment. As shown in fig. 6, the score prediction model training device 600 includes: a processor 602.
The processor 602 is configured to control and manage the actions of the score prediction model training apparatus, for example, execute the steps executed by the obtaining unit 501 and the processing unit 502, and/or execute other processes of the technical solutions described herein.
The processor 602 may be any means that can implement or execute the various illustrative logical blocks, modules, and circuits described in connection with the disclosure herein. The processor may be a central processing unit, general purpose processor, digital signal processor, application specific integrated circuit, field programmable gate array or other programmable logic device, transistor logic device, hardware component, or any combination thereof. Which may implement or execute the various illustrative logical blocks, modules, and circuits described in connection with the disclosure herein. The processor may also be a combination of computing functions, e.g., comprising one or more microprocessors, DSPs, and microprocessors, among others.
Optionally, the score prediction model training apparatus 600 may further include a communication interface 603, a memory 601, and a bus 604. The communication interface 603 is used to support the communication between the scoring prediction model training device 600 and other network entities. The memory 601 is used for storing the program codes and data of the scoring prediction model training device.
Wherein the memory 601 may be a memory in the score prediction model training apparatus, and the memory may include a volatile memory, such as a random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, a hard disk, or a solid state disk; the memory may also comprise a combination of memories of the kind described above.
The bus 604 may be an Extended Industry Standard Architecture (EISA) bus or the like. The bus 604 may be divided into an address bus, a data bus, a control bus, and the like. For ease of illustration, only one thick line is shown in FIG. 6, but this is not intended to represent only one bus or type of bus.
Through the above description of the embodiments, it is clear to those skilled in the art that, for convenience and simplicity of description, the foregoing division of the functional modules is merely used as an example, and in practical applications, the above function distribution may be completed by different functional modules according to needs, that is, the internal structure of the device may be divided into different functional modules to complete all or part of the above described functions. For the specific working processes of the system, the apparatus, and the module described above, reference may be made to the corresponding processes in the foregoing method embodiments, which are not described herein again.
The embodiment of the present application provides a computer program product containing instructions, which when run on an electronic device of the present application, causes the computer to execute the score prediction model training method described in the above method embodiment.
The embodiment of the present application further provides a computer-readable storage medium, where instructions are stored in the computer-readable storage medium, and when the computer executes the instructions, the electronic device of the present application executes each step executed by the score prediction model training apparatus in the method flows shown in the foregoing method embodiments.
The computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: an electrical connection having one or more wires, a portable computer diskette, and a hard disk. Random Access Memory (RAM), read-Only Memory (ROM), erasable Programmable Read-Only Memory (EPROM), registers, a hard disk, optical fiber, a portable Compact disk Read-Only Memory (CD-ROM), optical storage devices, magnetic storage devices, or any other form of computer-readable storage medium known in the art, in any suitable combination. An exemplary storage medium is coupled to the processor such the processor can read information from, and write information to, the storage medium. Of course, the storage medium may also be integral to the processor. The processor and the storage medium may reside in an Application Specific Integrated Circuit (ASIC). In embodiments of the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The above description is only an embodiment of the present application, but the scope of the present application is not limited thereto, and any changes or substitutions within the technical scope of the present disclosure should be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.
Claims (12)
1. A scoring prediction model training method, the method comprising:
acquiring sample data, wherein the sample data comprises multi-dimensional behavior characteristics of each user in a plurality of users and real user scores corresponding to the multi-dimensional behavior characteristics of each user;
determining a number of times each of the real user scores occurs in the sample data;
determining a target user score corresponding to the multi-dimensional behavior characteristics of each user according to the frequency of the score of each real user appearing in the sample data; the target user score is used for representing a predicted value of the score of each user;
and training the initial score prediction model according to the target user score corresponding to the multi-dimensional behavior characteristics of each user and the multi-dimensional behavior characteristics of each user to obtain a trained target score prediction model.
2. The method according to claim 1, wherein the determining the target user score corresponding to the multidimensional behavior feature of each user according to the number of times that each real user score appears in the sample data specifically comprises:
determining the distribution proportion of each real user score according to the frequency of each real user score appearing in the sample data;
rounding the initial prediction scores according to the distribution proportion of each real user score, and determining a target user score corresponding to the multi-dimensional behavior characteristic of each user; and the initial prediction score is the user score determined by the initial score prediction model.
3. The method of claim 2, wherein rounding the initial prediction scores according to the distribution ratio of each real user score comprises:
determining the distribution proportion of scores of every two real users with adjacent sizes;
according to the distribution proportion of each two real user scores adjacent in size, carrying out local rounding on the initial prediction score; according to the distribution proportion of each two real user scores adjacent in size, the initial prediction score is locally rounded to meet the following formula:
wherein A represents the target user score, B 1 And B 2 Representing each two real user scores adjacent in said magnitude, N representing said initial prediction score, x 1 And x 2 And the distribution proportion of each two real user scores adjacent to the size is represented.
4. The method of claim 3, wherein after the initial prediction score is locally rounded according to the distribution ratio of each two real user scores adjacent in size, the method further comprises:
determining the distribution proportion of any two real user scores;
according to the distribution proportion of any two real user scores, performing global rounding on the initial prediction score after the local rounding; wherein, according to the distribution ratio of any two real users, the global rounding of the initial prediction scores after the local rounding meets the following formula:
wherein A represents the target user score, B 3 And B 4 Representing the arbitrary two real user scores, N representing the initial predicted score, x 3 And x 4 Representing the distribution ratio of the arbitrary two real user scores.
5. The method according to any one of claims 1 to 4, wherein the multidimensional behavior characteristic of each user is determined based on the gender, age, number of times of using networks of different network standards, duration of using networks of different network standards, network call completing rate and average delay of Round Trip Time (RTT) of downlink.
6. A score prediction model training device, characterized by comprising: an acquisition unit and a processing unit;
the acquisition unit is used for acquiring sample data, wherein the sample data comprises the multidimensional behavior characteristics of each user in a plurality of users and the real user scores corresponding to the multidimensional behavior characteristics of each user;
the processing unit is used for determining the number of times each real user score appears in the sample data;
the processing unit is further configured to determine, according to the number of times that each real user score appears in the sample data, a target user score corresponding to the multidimensional behavior feature of each user; the target user score is used for representing a predicted value of the score of each user;
the processing unit is further configured to train the initial score prediction model according to the target user score corresponding to the multidimensional behavior feature of each user and the multidimensional behavior feature of each user, so as to obtain a trained target score prediction model.
7. A scoring prediction model training device as in claim 6,
the processing unit is further configured to determine a distribution ratio of each real user score according to the number of times that each real user score appears in the sample data;
the processing unit is further configured to round the initial prediction scores according to the distribution ratio of each real user score, and determine a target user score corresponding to the multidimensional behavior feature of each user; and the initial prediction score is the user score determined by the initial score prediction model.
8. The scoring prediction model training device according to claim 7,
the processing unit is also used for determining the distribution proportion of scores of every two real users with adjacent sizes;
the processing unit is further configured to perform local rounding on the initial prediction scores according to the distribution ratio of each two real user scores adjacent in size; according to the distribution proportion of each two real user scores adjacent in size, the initial prediction score is locally rounded to meet the following formula:
wherein A represents the target user score, B 1 And B 2 Representing each two real user scores adjacent in said magnitude, N representing said initial prediction score, x 1 And x 2 And the distribution proportion of each two real user scores adjacent to the size is represented.
9. The scoring prediction model training device according to claim 8,
the processing unit is also used for determining the distribution proportion of any two real user scores;
the processing unit is further configured to perform global rounding on the initial prediction score after the local rounding is performed according to a distribution ratio of any two real user scores; wherein, according to the distribution ratio of any two real users, the global rounding of the initial prediction score after the local rounding satisfies the following formula:
wherein A represents the target user score, B 3 And B 4 Representing the arbitrary two real user scores, N representing the initial predicted score, x 3 And x 4 Representing the distribution ratio of the arbitrary two real user scores.
10. The scoring prediction model training device according to any one of claims 6 to 9, wherein the multidimensional behavior characteristic of each user is determined based on the gender, age, the number of times of using networks of different network standards, the length of time of using networks of different network standards, a network call-through rate, and an average delay of a downlink round trip time RTT.
11. An electronic device, comprising: a processor and a memory; wherein the memory is configured to store computer-executable instructions that, when executed by the electronic device, are executed by the processor to cause the electronic device to perform the score prediction model training method of any one of claims 1-5.
12. A computer-readable storage medium comprising instructions that, when executed by an electronic device, enable the electronic device to perform the scoring predictive model training method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210836399.1A CN115203556A (en) | 2022-07-15 | 2022-07-15 | Score prediction model training method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210836399.1A CN115203556A (en) | 2022-07-15 | 2022-07-15 | Score prediction model training method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115203556A true CN115203556A (en) | 2022-10-18 |
Family
ID=83581388
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210836399.1A Pending CN115203556A (en) | 2022-07-15 | 2022-07-15 | Score prediction model training method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115203556A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116841914A (en) * | 2023-09-01 | 2023-10-03 | 星河视效科技(北京)有限公司 | Method, device, equipment and storage medium for calling rendering engine |
-
2022
- 2022-07-15 CN CN202210836399.1A patent/CN115203556A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116841914A (en) * | 2023-09-01 | 2023-10-03 | 星河视效科技(北京)有限公司 | Method, device, equipment and storage medium for calling rendering engine |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9684634B2 (en) | Method and apparatus for evaluating predictive model | |
| CN108683530B (en) | Data analysis method and device for multi-dimensional data and storage medium | |
| CN110489449B (en) | Chart recommendation method and device and electronic equipment | |
| CN109165691B (en) | Training method and device for model for identifying cheating users and electronic equipment | |
| CN108388509B (en) | Software testing method, computer readable storage medium and terminal equipment | |
| CN110598109A (en) | Information recommendation method, device, equipment and storage medium | |
| CN111797320A (en) | Data processing method, device, equipment and storage medium | |
| CN112580896A (en) | Knowledge point prediction method, knowledge point prediction device, knowledge point prediction equipment and storage medium | |
| CN111639493A (en) | Address information standardization method, device, equipment and readable storage medium | |
| CN111178537A (en) | Feature extraction model training method and device | |
| CN114428748B (en) | Simulation test method and system for real service scene | |
| CN115203556A (en) | Score prediction model training method and device, electronic equipment and storage medium | |
| CN114692889A (en) | Meta-feature training model for machine learning algorithm | |
| CN111784246B (en) | Logistics path estimation method | |
| CN113761193A (en) | Log classification method and device, computer equipment and storage medium | |
| CN111783883A (en) | Abnormal data detection method and device | |
| CN114911814A (en) | Consultation service method and system based on knowledge resource library updating | |
| CN112783747B (en) | Execution time prediction method and device for application program | |
| CN110134575B (en) | Method and device for calculating service capacity of server cluster | |
| CN112613762A (en) | Knowledge graph-based group rating method and device and electronic equipment | |
| CN117744547B (en) | Method and device for predicting circuit device resources, electronic equipment and storage medium | |
| CN110309127B (en) | Data processing method and device and electronic equipment | |
| CN109299349B (en) | Application recommendation method and device, equipment and computer-readable storage medium | |
| CN115345637A (en) | Consumption prediction method, consumption prediction device, electronic equipment and computer readable storage medium | |
| CN115828911A (en) | Test question knowledge point identification method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |