Prediction model generation method and device and computer readable storage medium
By generating predictive models through selecting functional configuration packages from the recipe library, the problem of high costs in marketing forecasting solutions is solved, enabling rapid, low-cost model training and automatic iterative upgrades to adapt to enterprise business needs.
Patent Information
- Application Number
- CN202411394105.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-08
- Publication Date
- 2024-11-08
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing marketing forecasting solutions are costly, rely on professional teams, and result in long and inefficient AI implementation cycles for enterprises, as they cannot automatically adapt to model iteration and upgrades.
By pre-setting a recipe library, functional configuration packages can be retrieved from it based on input information to generate predictive models that meet business characteristics, including the encapsulation of datasets, hyperparameters, and model code, simplifying the training process and reducing the involvement of technical professionals.
It enables the rapid generation of predictive models tailored to specific business characteristics, reducing costs, simplifying the implementation process, automatically adapting to model iterations and upgrades, and improving predictive performance.
Smart Images

Figure CN118917882A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of data processing technology, and in particular to a prediction model generation method and device, and a computer-readable storage medium. Background Art
[0002] At present, due to the development of data science, prediction in the field of intelligent marketing is also known as one of the emerging fields. Intelligent marketing prediction usually requires professional team members to develop according to daily business.
[0003] In the prior art, when users conduct marketing forecasting for daily business, they need to establish their own expert development team or entrust external professionals, such as third-party outsourced project service providers.
[0004] However, the cost of marketing forecasting solutions in the prior art is relatively high. Summary of the invention
[0005] The present application provides a solution that can quickly train a prediction model suitable for its own business characteristics, thereby reducing the cost of generating the prediction model.
[0006] In order to achieve the above objectives, this application provides the following technical solutions: In a first aspect, a prediction model generation method is provided, the prediction model generation method comprising: retrieving a corresponding first function configuration package from a recipe library according to input information, the first function configuration package being used to configure relevant parameters of the prediction model; acquiring first source data according to the first function configuration package; processing the first source data according to the first function configuration package, and generating a first prediction model using the processed first source data.
[0007] Optionally, the recipe library includes multiple candidate function configuration packages, each candidate function configuration package is used to configure at least one of the following parameters: a data set, a hyperparameter of a prediction model, and a function of the prediction model.
[0008] Optionally, each candidate function configuration package includes at least one of the following: a definition of the data set, a processing flow of the data set, hyperparameters of the prediction model, a code of the prediction model, and an evaluation index of the prediction model.
[0009] Optionally, processing the first source data according to the first functional configuration package includes: dividing the processed first source data into a training set and a validation set; generating the first prediction model using the processed first source data includes: training the first prediction model using the training set; and validating the trained first prediction model using the validation set.
[0010] Optionally, processing the first source data according to the first functional configuration package includes: performing at least one of the following data processing on the first source data: data cleaning, feature extraction, and normalization.
[0011] Optionally, processing the first source data according to the first functional configuration package includes: performing quality analysis on the first source data, and outputting prompt information or modifying the first source data according to the quality analysis result.
[0012] Optionally, the prediction model generation method also includes: generating a second prediction model using the first functional configuration and the second source data, or generating the second prediction model using the second functional configuration and the first source data; and performing prediction using the first prediction model and the second prediction model respectively.
[0013] Optionally, the prediction model generation method also includes: deploying the first prediction model as an application program interface.
[0014] In a second aspect, the present application also discloses a prediction model generation device, which includes: a calling module, used to call a corresponding first function configuration package from a recipe library according to input information, wherein the first function configuration package is used to configure relevant parameters of the prediction model; an acquisition module, used to acquire first source data according to the first function configuration package; and a model generation module, used to process the first source data according to the first function configuration package, and generate a first prediction model using the processed first source data.
[0015] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. The computer program is executed by a processor to execute a method provided in the first aspect.
[0016] In a fourth aspect, a terminal device is provided, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the computer program to execute a method provided in the first aspect.
[0017] According to a fifth aspect, a computer program product is provided, on which a computer program is stored, and the computer program is executed by a processor to execute a method provided in the first aspect.
[0018] In a sixth aspect, an embodiment of the present application further provides a chip (or a data transmission device) on which a computer program is stored, and when the computer program is executed by the chip, the steps of the above method are implemented.
[0019] In the seventh aspect, an embodiment of the present application also provides a system chip, which is applied to a terminal device. The chip system includes at least one processor and an interface circuit. The interface circuit and the at least one processor are interconnected through lines. The at least one processor is used to execute instructions to execute a method provided in the first aspect.
[0020] Compared with the prior art, the technical solution of this application has the following beneficial effects: In the technical solution of the present application, by pre-setting the recipe library, the corresponding first function configuration package is retrieved from the recipe library according to the input information, and the first prediction model is generated according to the first function configuration package and the first source data. In the technical solution of the present application, the user only needs to select the appropriate first function configuration package in the recipe library to automatically generate a prediction model that meets the requirements, thereby achieving the purpose of quickly training a prediction model suitable for its own business characteristics; in addition, the prediction model generation scheme is simple and easy to implement, does not require excessive participation of professional and technical personnel, and can also reduce the generation cost of the prediction model.
[0021] Furthermore, the present application can also perform quality analysis on the first source data to determine the quality of the first source data, such as the proportion of dirty data, the proportion of null value data, etc.; and modify the first source data based on the quality analysis results to ensure the training effect of the subsequent prediction model. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a structural schematic diagram of a prediction model generation system provided in an embodiment of the present application; Figure 2 is a flow chart of a prediction model generation method provided in an embodiment of the present application; Figure 3 It is a structural schematic diagram of a formula library provided in an embodiment of the present application; Figure 4 is a schematic diagram of a design module provided in an embodiment of the present application; Figure 5 is a schematic diagram of a service module provided in an embodiment of the present application; Figure 6 is a flowchart of another prediction model generation method provided in an embodiment of the present application; Figure 7 It is a structural schematic diagram of a prediction model generating device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0023] As described in the background art, the cost of marketing forecasting solutions in the prior art is relatively high.
[0024] Specifically, the existing technology relies on data scientists or data engineers to complete the training and configuration process of marketing prediction models. The defects of this method are very obvious. First, it cannot effectively reduce the cost of enterprise application of artificial intelligence (AI); second, the AI implementation cycle is very long and inefficient.
[0025] In addition, existing technologies cannot automatically adapt to the requirements of model iteration and upgrading, which often leads to a decrease in the prediction effect of the model after a period of time due to the inability to cover some new behaviors.
[0026] In the technical solution of the present application, by pre-setting the recipe library, the corresponding first function configuration package is retrieved from the recipe library according to the input information, and the first prediction model is generated according to the first function configuration package (also called the algorithm configuration package) and the first source data. In the technical solution of the present application, the user only needs to select the appropriate first function configuration package in the recipe library to automatically generate a prediction model that meets the requirements, thereby achieving the purpose of quickly training a prediction model suitable for its own business characteristics; in addition, the prediction model generation scheme is simple and easy to implement, does not require excessive participation of professional and technical personnel, and can also reduce the generation cost of the prediction model.
[0027] The prediction model in this application can be used for business prediction in consumption scenarios, such as whether consumers will churn and the probability of churn; the probability of consumers using consumption coupons when consumer coupons are issued to consumers; the probability that consumers are price-sensitive groups; the prediction of consumers' next purchase of goods, etc. Different businesses correspond to different prediction models.
[0028] The platform used by the technical solution of this application can be an intelligent decision-making (AI Decision) platform, which allows users to quickly implement their business data and experience into marketing prediction models and apply them to actual operational business. The technical solution of this application can assist the decision-making of the operation team and empower the operational business process of the enterprise. By encapsulating specific functional configuration packages, for example, encapsulating algorithms, data models, data flow schemes, hyperparameters, training codes, evaluation codes, evaluation labels, etc. for specific scenarios into functional configuration packages, training can be automatically completed through structured enterprise data, and intelligent services can be quickly applied to business scenarios.
[0029] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below with reference to the accompanying drawings.
[0030] See also Figure 1 , Figure 1 A schematic diagram of a prediction model generation system of the present application is shown.
[0031] The prediction model generation system 10 includes a recipe library 101, a design module 102 and a service module 103. The recipe library 101 includes multiple candidate function configuration packages, which can also be called recipes, algorithm configuration packages, etc. Each function configuration package can be a pre-packaged algorithm module. The multiple function configuration packages in the recipe library 101 can cover different business scenarios.
[0032] In a specific implementation, each candidate function configuration package is used to configure at least one of the following parameters: Dataset, hyperparameters of the prediction model, and functions of the prediction model. The functions of the prediction model can specifically include predicting purchase preferences, predicting the timing of repurchase, predicting the timing of the next interaction, predicting coupon sensitivity, etc.
[0033] Specifically, each candidate function configuration package may include at least one of the following: Dataset definition is used to define a data set, which can specifically point to at least one data table / data structure in a business system, that is, from which data table or data structure to obtain data; The processing flow of the data set may include the data set feature processing process, such as data cleaning, feature extraction, normalization, etc., to normalize the data, which is conducive to the subsequent model processing of data; it may also include the division and configuration of the training set and the validation set, for example, 80% of the data set is the training set and 20% is the validation set; Configuration items of hyperparameters of prediction models; Prediction model code, such as code for model training, code for model evaluation, code for model prediction, etc.; Evaluation metrics used to evaluate predictive models.
[0034] For a user, at least one first function configuration package can be selected from a plurality of candidate function configuration packages in the recipe library 101 , where the first function configuration package is one or more candidate function configurations in the plurality of candidate function configuration packages.
[0035] Continue to refer to Figure 1 The design module 102 generates a prediction model according to at least one first function configuration package, specifically generates a prediction model according to the first function configuration package selected by the user, such as a training prediction model and a verification prediction model.
[0036] The service module 103 publishes the generated prediction model as an application programming interface (API). The prediction model will provide prediction services to users through the API interface.
[0037] In a specific implementation, the API interface of the prediction model can be provided to the business system, and the business system uses the prediction model to perform business prediction through the API interface. Alternatively, the API interface of the prediction model can be directly provided to the user, and the user can directly call the API interface to use the prediction model to perform business prediction.
[0038] See also Figure 2 , Figure 2 A prediction model generation method is disclosed. Figure 2 The prediction model generation method shown can be applied to a computer device, which may include a server, a desktop computer, etc. Alternatively, the prediction model generation method can be applied to other appropriate entities, which is not limited in the present application.
[0039] It is understandable that, in a specific implementation, the prediction model generation method can be implemented in the form of a software program, and the software program runs in a processor integrated inside a chip or a chip module. The method can also be implemented in the form of software combined with hardware, which is not limited in this application.
[0040] The method provided in this application specifically includes steps 201 to 203.
[0041] In step 201, a corresponding first function configuration package is retrieved from a recipe library according to input information, wherein the first function configuration package is used to configure relevant parameters of a prediction model.
[0042] In a specific implementation, the user may input or select input information in the user interface, and based on the input information, the corresponding first functional configuration package may be retrieved from the recipe library. Specifically, the input information may point to a specific functional configuration package, or the semantics of the input information may match the semantics of the first functional configuration package.
[0043] Specifically, continue to refer to Figure 1 , the user can select one or more candidate function configuration packages from multiple candidate function configuration packages in the recipe library 101 as the first function configuration package.
[0044] For example, if the user selects the first function configuration package “purchase preference”, then the first function configuration package “purchase preference” will be used to construct a prediction model for predicting consumer purchase preference.
[0045] Specifically, the recipe library may include candidate function configuration packages for various business scenarios. Figure 3As shown, the candidate function configuration packages in the recipe library can be divided into four parts according to business scenarios: consumption, interaction, promotion and identity. Among them, the consumption scenario can include candidate function configuration packages such as "purchase preference", "potential churn prediction", "life cycle value", "repurchase timing prediction", and "first order high conversion". Correspondingly, the interaction scenario can include candidate function configuration packages such as "next interaction time", "best contact method", "best contact time", and "best contact channel". The interaction scenario can include candidate function configuration packages such as "coupon sensitivity", "discount promotion sensitivity", "activity participation", and "price sensitivity". The identity scenario can include candidate function configuration packages such as "price sensitive group identification", "consumer gender", "age group", "income level", and "recent consumption time, consumption frequency, and consumption amount (Recency, Frequency, Monetary, RFM)".
[0046] It should be noted that the specific functions of the candidate function configuration packages in the recipe library can be adaptively set according to the actual application scenarios, and this application does not impose any restrictions on this.
[0047] In step 202, first source data is acquired according to a first function configuration package.
[0048] In a specific implementation, the source of data can be defined in the first function configuration package, so the first source data can be obtained according to the first function configuration package, and the first source data will be used to train and verify the prediction model. Since the data required to generate prediction models for different businesses are different, it is necessary to define the data in the first function configuration package, and when generating the prediction model, the corresponding first source data is obtained from the business system according to the requirements of the first function configuration package.
[0049] Specifically, the first function configuration package may include a definition of a data set, which is used to define a data set, and may specifically point to at least one data table / data structure in a business system, that is, from which data table or data structure to obtain data. Thus, the first source data may be obtained from the pointed data table / data structure.
[0050] In step 203, the first source data is processed according to the first functional configuration package, and a first prediction model is generated using the processed first source data.
[0051] In a specific implementation, in order to better train the prediction model, the first source data may be normalized, and the processed first source data will be used to train and verify the prediction model.
[0052] In some non-limiting embodiments, since the processing flow of the data set is defined in the first functional configuration package, the data set feature processing process can be performed on the first source data according to the first functional configuration package. The data set feature processing process can specifically include at least one of the following data processing: data cleaning, feature extraction, and normalization.
[0053] Among them, by performing data cleaning on the first source data, dirty data, empty value data, etc. in the first source data can be removed, valid data can be retained, and the effect of prediction model training can be improved. By performing feature extraction on the first source data, features in the first source data can be extracted, and the features will be used to generate the model to improve the efficiency of model generation. By performing normalization processing on the first source data, the data can be standardized to avoid the influence of noise data on model training, and the model training effect can be further improved.
[0054] In some non-limiting embodiments, the first functional configuration package defines the division configuration of the training set and the validation set, so the processed first source data can be divided into the training set and the validation set according to the first functional configuration package. For example, if the first functional configuration package defines 80% of the data set as the training set and 20% as the validation set, then 80% of the first source data can be divided into the training set and 20% of the first source data can be divided into the validation set.
[0055] Furthermore, the training set is used to train the first prediction model; and the validation set is used to validate the trained first prediction model.
[0056] The embodiment of the present application sets up a recipe library in advance, retrieves the corresponding first function configuration package from the recipe library according to the input information, and generates a first prediction model according to the first function configuration package (also called an algorithm configuration package) and the first source data. In the technical solution of the present application, the user only needs to select a suitable first function configuration package in the recipe library to automatically generate a prediction model that meets the requirements, thereby achieving the purpose of quickly training a prediction model suitable for its own business characteristics; in addition, the prediction model generation scheme is simple and easy to implement, does not require excessive participation of professional and technical personnel, and can also reduce the generation cost of the prediction model.
[0057] In some non-limiting embodiments, please refer to Figure 1 and Figure 4 , which can be executed by the design module 102 in the prediction model generation system 10 Figure 2 Steps 102 and 103 are shown to generate a first prediction model.
[0058] In a specific implementation, the first prediction model can be generated by the model management unit 1021 in the design module 102. The model training function of the model management unit 1021 can be used to train the first prediction model; the model training log function can generate a log of the process of training the first prediction model for subsequent viewing; the model evaluation function can evaluate the first prediction model according to the evaluation indicators in the first function configuration package; the model evaluation report function can generate a model evaluation report according to the evaluation results; the model publishing function can be used to publish the first prediction model, for example, publishing the first prediction model as an API interface.
[0059] Specifically, the model creation wizard function can allow users to create a task for training a first prediction model. For example, the task indicates which model to train and how often to train.
[0060] In a specific implementation, the training of the first prediction model can be performed periodically, that is, the first prediction model is retrained using new business data at regular intervals. The training scheduling function can compare the performance of the prediction models obtained from the two previous and subsequent trainings.
[0061] In a specific embodiment, the second prediction model is generated using the first functional configuration and the second source data, and the first prediction model and the second prediction model are used to perform predictions respectively.
[0062] In this embodiment, the second source data is different from the first source data. For example, the first source data is business data for 2023, and the second source data is business data for 2024. The prediction performance of prediction models generated using different source data may be different. In order to provide better prediction services, the first prediction model and the second prediction model can be used for prediction respectively.
[0063] Further, the performance of the first prediction model and the second prediction model is determined in combination with the actual data according to the first prediction result of the first prediction model and the second prediction result of the second prediction model. For example, if the first difference between the first prediction result and the actual data is greater than the second difference between the second prediction result and the actual data, then it can be determined that the performance of the second prediction model is better, otherwise, it can be determined that the performance of the first prediction model is better.
[0064] The embodiment of the present application uses prediction models generated by different source numbers and performs AB testing, which can automatically adapt to the requirements of model iteration and upgrading, so that the prediction model can cover some new behaviors after a period of time to ensure the prediction effect.
[0065] In addition to automatically generating the first prediction model, the model management unit 1021 also provides an external model import function, through which the user can import an external model.
[0066] Continue to refer to Figure 4 The platform management unit 1022 in the design module 102 is used to manage the computing power platform and resource space. Specifically, the platform configuration management function can be used to select a platform from the platforms that provide computing power, such as Alibaba Cloud, Microsoft, etc., for the subsequent generation of prediction models; the workspace information function is used to manage the allocated resource space.
[0067] The scenario management unit 1023 in the design module 102 is used to manage specific business scenarios. Specifically, the preset scenario solution function is used to view, enable, and disable preset scenario solutions, wherein each scenario solution corresponds to a functional configuration package with more descriptions to help operators understand. The associated data set function is used to associate a specific business scenario with a specific data set, which includes specific data. When the prediction model is subsequently generated, the associated data set can be directly called. Specifically, the first source data, such as a data table, can be extracted according to the first functional configuration to form a data set, which is stored to form a data set associated with the business scenario. The solution parameter configuration function can be used to modify the parameters in the functional configuration package. Generally speaking, the parameter in the first functional configuration is the default value. The value of the parameter can be modified, or the default value in the first functional configuration can be selected.
[0068] The data set management unit 1024 in the design module 102 can be used to manage specific data sets. Among them, the data set analysis function can be used to perform quality analysis on the first source data, and output prompt information according to the quality analysis results, such as prompting the user to delete; or modify the first source data, such as deleting 1,000 pieces of data with poor quality. This is because the data table structure in the source of the first source data is consistent but the data quality may be different, such as more dirty data and more null values, so the data quality can be improved through data set analysis.
[0069] The present application can also perform quality analysis on the first source data to determine the quality of the first source data, such as the proportion of dirty data, the proportion of null value data, etc.; and modify the first source data based on the quality analysis results to ensure the training effect of the subsequent prediction model.
[0070] In addition, the dataset maintenance function can be used to modify and edit the dataset. The dataset publishing function can be used to publish the verified data, which can be used for subsequent selection as training data or verification data. Among them, the verified data can be data that meets the quality requirements after the dataset analysis.
[0071] In the embodiment of the present application, through the formula library 101 and the design module 102, the user can use his own business data for migration training to obtain a scenario-based prediction model that is more suitable for his own business.
[0072] In some non-limiting embodiments, reference is made to Figure 1 and Figure 5 The service module 103 centrally manages the intelligent services and implements intelligent prediction in the form of an API interface, specifically supporting batch processing mode and real-time prediction mode. In other words, the service module 103 is a standardized service interface between the design module 102 and the business system.
[0073] Among them, the model service deployment function is used to deploy the prediction model as an API. The service verification function is used to debug and test the API to verify whether the API meets the requirements. The service details function is used to record the situation of each service provision. The service log is used to record the use of each API interface. The service monitoring information is used to monitor and record abnormal data. The service AB test deployment function is used to compare the performance of two prediction models. Specifically, tasks can be allocated to the two prediction models in proportion, and services can be provided using the two prediction models. The better prediction model can be determined based on the log. The two prediction models can be trained using different source data, or generated using parameters in different function configuration packages.
[0074] Specifically, the two prediction models can be a first prediction model and a second prediction model, the first prediction model is generated based on the first function configuration package and the first source data, the second prediction model is generated based on the first function configuration package and the second source data, or the second prediction model is generated based on the second function configuration package and the first source data.
[0075] Please refer to Figure 6 , Figure 6 A specific prediction model generation process is shown. The prediction model generated by this process has better performance and higher efficiency.
[0076] In step 601, a corresponding first function configuration package is retrieved from a recipe library according to input information.
[0077] In step 602, first source data is acquired according to the first function configuration package.
[0078] In step 603, at least one of the following data processing is performed on the first source data: data cleaning, feature extraction, and normalization.
[0079] The embodiment of the present application performs normalization on the first source data, which facilitates subsequent model processing of the data.
[0080] In step 604, the processed first source data is divided into a training set and a validation set.
[0081] In step 605, the first prediction model is trained using the training set, and the trained first prediction model is verified using the verification set.
[0082] In step 606, the first prediction model is deployed as an application program interface.
[0083] In the embodiment of the present application, the first prediction model will provide prediction services to users through the API interface. The business system can use the first prediction model to perform business prediction by calling the API interface.
[0084] In a specific application scenario, the user needs to build a prediction model for predicting the next purchase timing of a consumer. The user selects the corresponding first functional configuration package in the recipe library, such as "repurchase timing prediction". The design module will obtain the corresponding business data, that is, the first source data, based on the first functional configuration package, and use the first source data to generate a first prediction model. The first prediction model is published as an API interface. The user's business system can predict the next purchase timing of a consumer by calling the API interface.
[0085] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.
[0086] Please refer to Figure 7 , Figure 7 A prediction model generating device 70 is shown, and the prediction model generating device 70 may include: The retrieval module 701 is used to retrieve the corresponding first function configuration package from the recipe library according to the input information, where the first function configuration package is used to configure the relevant parameters of the prediction model; An acquisition module 702 is used to acquire first source data according to a first function configuration package; The model generation module 703 is used to process the first source data according to the first function configuration package, and generate a first prediction model using the processed first source data.
[0087] In the embodiment of the present application, by pre-setting the recipe library, the corresponding first function configuration package is retrieved from the recipe library according to the input information, and the first prediction model is generated according to the first function configuration package and the first source data. In the technical solution of the present application, the user only needs to select the appropriate first function configuration package in the recipe library to automatically generate a prediction model that meets the requirements, thereby achieving the purpose of quickly training a prediction model suitable for its own business characteristics; in addition, the prediction model generation scheme is simple and easy to implement, does not require excessive participation of professional and technical personnel, and can also reduce the generation cost of the prediction model.
[0088] It should be noted that the acquisition module 702 can implement the same function as the data set management unit 1024 mentioned in the above embodiment; the model generation module 703 can implement the same function as the model set management unit 1021 mentioned in the above embodiment. The specific implementation of the acquisition module 702 and the model generation module 703 can refer to the above embodiments of the data set management unit 1024 and the model set management unit 1021, and this application will not repeat them here.
[0089] In a specific implementation, the above-mentioned communication device 70 may correspond to a chip in a terminal device having a prediction model generation function, such as an SOC, a baseband chip, etc.; or correspond to a chip module in a terminal device including a prediction model generation function; or correspond to a chip module having a data processing function chip, or correspond to a network device.
[0090] For other relevant descriptions about the prediction model generating device 70, reference may be made to the relevant descriptions in the aforementioned embodiments, which will not be repeated here.
[0091] Regarding the various modules / units included in the various devices and products described in the above embodiments, they can be software modules / units, or hardware modules / units, or they can be partially software modules / units and partially hardware modules / units. For example, for various devices and products applied to or integrated in a chip, the various modules / units included therein can all be implemented in the form of hardware such as circuits, or at least some of the modules / units can be implemented in the form of software programs, which run on the processor integrated inside the chip, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits; for various devices and products applied to or integrated in a chip module, the various modules / units included therein can all be implemented in the form of hardware such as circuits, and different modules / units can be located in the same component of the chip module (such as a chip, circuit module, etc.) or in different components, or at least some of the modules / units can be implemented in the form of software programs. The software programs run on the processor integrated inside the chip, and the remaining (if any) modules / units can be implemented in the form of hardware such as circuits; for various devices and products applied to or integrated in a chip module, the various modules / units included therein can all be implemented in the form of hardware such as circuits, and different modules / units can be located in the same component of the chip module (such as a chip, circuit module, etc.) or in different components, or at least some of the modules / units can be It is implemented in the form of a software program that runs on a processor integrated inside the chip module, and the remaining (if any) modules / units can be implemented in hardware such as circuits; for various devices and products applied to or integrated in the terminal equipment, the various modules / units contained therein can be implemented in hardware such as circuits, and different modules / units can be located in the same component (for example, chip, circuit module, etc.) or different components in the terminal equipment, or at least some of the modules / units can be implemented in the form of a software program that runs on a processor integrated inside the terminal equipment, and the remaining (if any) modules / units can be implemented in hardware such as circuits.
[0092] The embodiment of the present application also discloses a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is run, the steps of the method shown in the above embodiment can be executed. The storage medium may include a read-only memory (ROM), a random access memory (RAM), a disk or an optical disk, etc. The storage medium may also include a non-volatile memory (non-volatile) or a non-transitory memory, etc.
[0093] It should be understood that the term "and / or" in this article is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article indicates that the associated objects before and after are in an "or" relationship.
[0094] The "plurality" appearing in the embodiments of the present application refers to two or more.
[0095] The first, second, etc. descriptions appearing in the embodiments of the present application are only used for illustration and distinction of the description objects. There is no order, nor do they indicate any special limitation on the number of devices in the embodiments of the present application, and cannot constitute any limitation on the embodiments of the present application.
[0096] The "connection" that appears in the embodiments of the present application refers to various connection methods such as direct connection or indirect connection to achieve communication between devices, and the embodiments of the present application do not impose any limitations on this.
[0097] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented using software, the above embodiments may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless means.
[0098] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0099] In the several embodiments provided in the present application, it should be understood that the disclosed methods, devices and systems can be implemented in other ways. For example, the device embodiments described above are merely schematic; for example, the division of the units is only a logical function division, and there may be other division methods in actual implementation; for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.
[0100] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0101] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may be physically included separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional units.
[0102] The above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium. The above-mentioned software functional unit is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute some steps of the method described in each embodiment of the present application.
[0103] Although the present application is disclosed as above, the present application is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application shall be subject to the scope defined by the claims.
Claims
1. A prediction model generation method, It is characterized in that include: Retrieving a corresponding first function configuration package from a recipe library according to the input information, wherein the first function configuration package is used to configure relevant parameters of the prediction model; Acquire first source data according to the first function configuration package; The first source data is processed according to the first functional configuration package, and a first prediction model is generated using the processed first source data.
2. The prediction model generation method according to claim 1, It is characterized in that The recipe library includes a plurality of candidate function configuration packages, each of which is used to configure at least one of the following parameters: The dataset, the hyperparameters of the predictive model, and the functionality of the predictive model.
3. The prediction model generation method according to claim 2, It is characterized in that Each candidate functional configuration package includes at least one of the following: The definition of the data set, the processing flow of the data set, the hyperparameters of the prediction model, the code of the prediction model, and the evaluation indicators of the prediction model.
4. The prediction model generation method according to claim 1, It is characterized in that The processing of the first source data according to the first function configuration package comprises: Dividing the processed first source data into a training set and a validation set; The generating the first prediction model by using the processed first source data comprises: Using the training set to train the first prediction model; The trained first prediction model is validated using the validation set.
5. The prediction model generation method according to claim 1, It is characterized in that The processing of the first source data according to the first function configuration package comprises: Perform at least one of the following data processing on the first source data: data cleaning, feature extraction, and normalization.
6. The prediction model generation method according to claim 1, It is characterized in that The processing of the first source data according to the first function configuration package comprises: Perform a quality analysis on the first source data, and output prompt information or modify the first source data according to the quality analysis result.
7. The prediction model generation method according to claim 1, It is characterized in that Also includes: Generate a second prediction model using the first functional configuration and the second source data, or generate the second prediction model using the second functional configuration and the first source data; The first prediction model and the second prediction model are used to perform prediction respectively.
8. The prediction model generation method according to claim 1, It is characterized in that Also includes: The first prediction model is deployed as an application program interface.
9. A prediction model generating device, It is characterized in that include: A calling module, used to call a corresponding first function configuration package from a recipe library according to input information, wherein the first function configuration package is used to configure relevant parameters of the prediction model; An acquisition module, configured to acquire first source data according to the first function configuration package; A model generation module is used to process the first source data according to the first function configuration package, and generate a first prediction model using the processed first source data.
10. A computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the prediction model generation method according to any one of claims 1 to 8 are executed.
11. A computer program product comprising a computer program / instructions, It is characterized in that When the computer program / instructions are executed by a processor, the steps of the prediction model generation method described in any one of claims 1 to 8 are implemented.
12. A terminal device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, It is characterized in that When the processor runs the computer program, the processor performs the steps of the prediction model generation method according to any one of claims 1 to 8.
Citation Information
Patent Citations
Model generation method and device, electronic equipment and storage medium
CN111310934A
Competitive product enterprise matching and information standardization method based on pre-training model prompt project
CN117034907A
Program editor for integrating AI into PLC and integration method
CN117724390A
Marketing effect prediction method and device, electronic equipment and medium
CN118411199A
Generation method and device of machine learning model, equipment and medium
CN118674075A
