A business data processing method, device, electronic device and medium

Abstract

The present application provides a business data processing method, device, electronic device and medium, the method comprising: obtaining target business data from multiple different data sources; sending the target business data to multiple types of data processing servers; different types of data processing servers process the target business data based on configured data processing rules, respectively determine the target business feature data that matches the types of the different types of AI model servers, and send them to AI model servers of matching types; different types of AI model servers process the received business feature data based on pre-configured and trained business volume prediction models, and determine different types of business volume prediction results; and fuse the different types of business volume prediction results to determine the target business volume prediction result, thereby improving the data processing speed and the real-time level and prediction accuracy of business volume prediction.

Description

Technical Field

[0001] The present application relates to the field of big data processing technology, and more specifically, to a business data processing method, device, electronic device and medium. Background Art

[0002] As the scale of enterprises expands and market competition intensifies, in the context of the big data era, the amount of data generated by enterprises every day is growing exponentially, covering multiple dimensions such as sales, production, supply chain, and market feedback. Although the existing financial forecasting system has designed some forecasting algorithms to keep up with the times, it is often difficult to quickly process these massive amounts of data due to outdated architecture design and inefficient algorithms. As a result, the existing financial forecasting technology is limited by the slow processing speed and insufficient accuracy of business data, and is unable to provide decision support information in a timely manner. This lag makes it difficult for enterprises to respond quickly and accurately when facing market opportunities or risks. Summary of the invention

[0003] In view of this, the purpose of the present application is to provide a business data processing method, device, electronic device and medium to improve the data processing speed and the real-time level and prediction accuracy of business volume prediction.

[0004] An embodiment of the present application provides a method for processing business data, the method comprising:

[0005] Acquire target business data from multiple different data sources;

[0006] Sending the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers;

[0007] The different types of data processing servers process the target business data based on the configured data processing rules, respectively, determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type;

[0008] The different types of AI model servers process the received service feature data based on pre-configured and trained service volume prediction models to determine different types of service volume prediction results;

[0009] The different types of business volume forecast results are integrated to determine a target business volume forecast result.

[0010] In some embodiments, in the business data processing method, after determining the target business volume prediction result, the method further includes:

[0011] Acquire actual business data corresponding to the target business volume prediction result from a plurality of different data sources;

[0012] Processing the actual business data, generating sample data corresponding to prediction models in different types of AI model servers, and obtaining different types of sample data;

[0013] The traffic volume prediction model is trained based on type-matched sample data obtained in a preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step.

[0014] In some embodiments, in the business data processing method, the training of the business volume prediction model based on the type-matched sample data obtained in the preset time step to update the business volume prediction model in the AI ​​model server according to the preset time step includes:

[0015] The traffic volume prediction model processes the sample data of the type matching received in the preset time step to determine the loss result of the loss function of the traffic volume prediction model under the current parameters;

[0016] Calculate the gradient of the loss function with respect to the current parameter based on the loss result;

[0017] Based on the gradient, and the pre-configured update rules and learning rate of the gradient descent method, the parameters of the business volume prediction model are updated to update the business volume prediction model in the AI ​​model server according to a preset time step.

[0018] In some embodiments, in the business data processing method, after acquiring actual business data corresponding to the target business volume prediction result from a plurality of different data sources, the method further includes:

[0019] Comparison results of comparing the actual traffic volume in the actual traffic data with traffic volume prediction results of traffic volume prediction models in different types of AI model servers;

[0020] Based on the comparison result, a feedback result for the traffic volume prediction model in the AI ​​model server is generated.

[0021] In some embodiments, in the business data processing method, the different types of AI model servers process the received business feature data based on pre-configured and trained business volume prediction models to determine different types of business volume prediction results, including:

[0022] Each type of AI model server is pre-configured with multiple trained business volume prediction models, and the multiple business volume prediction models respectively process the received business feature data to determine multiple business volume prediction results;

[0023] The multiple business volume prediction results are integrated to determine the business volume prediction result of this type of AI model server.

[0024] In some embodiments, in the business data processing method, the different types of AI model servers include a first AI model server for sequential business data and a second AI model server for non-sequential business data;

[0025] The traffic volume prediction model in the first AI model server includes a first traffic volume prediction model obtained by training based on the LSTM model and a second traffic volume prediction model obtained by training based on the GRU model;

[0026] The traffic volume prediction model in the second AI model server includes a third traffic volume prediction model trained based on a random forest model and a fourth traffic volume prediction model trained based on a linear regression model.

[0027] In some embodiments, in the business data processing method, after determining the target business volume prediction result, the method further includes:

[0028] The target traffic volume prediction result is processed into a chart and / or a report, and the chart and / or the report is sent to a terminal device so as to display the target traffic volume prediction result on a graphical user interface of the terminal device.

[0029] In some embodiments, a business data processing device is further provided, the device comprising:

[0030] An acquisition module is used to acquire target business data from multiple different data sources;

[0031] A sending module, used to send the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers;

[0032] A first processing module is used to enable the different types of data processing servers to process the target business data based on the configured data processing rules, respectively, determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type;

[0033] A determination module, configured to enable the different types of AI model servers to process the received service feature data based on pre-configured and trained service volume prediction models, respectively, to determine different types of service volume prediction results;

[0034] The fusion module is used to fuse the different types of business volume prediction results to determine the target business volume prediction result.

[0035] In some embodiments, an electronic device is also provided, comprising: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate through the bus, and when the machine-readable instructions are executed by the processor, the steps of the business data processing method are performed.

[0036] In some embodiments, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of the business data processing method are executed.

[0037] The embodiments of the present application provide a business data processing method, device, electronic device and medium, the method comprising obtaining target business data from multiple different data sources; sending the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model server types; the different types of data processing servers respectively process the target business data based on the configured data processing rules, respectively determine the target business feature data that matches the different types of AI model server types, and send the target business feature data to the matching AI model server; the different types of AI model servers respectively process the received business feature data based on a pre-configured and trained business volume prediction model; Collect data to determine different types of business volume prediction results; fuse the different types of business volume prediction results to determine the target business volume prediction results; the method can instantly capture, clean, convert and predict data streams from multiple sources; various types of data processing servers are configured with data processing rules that match specific types of AI model servers, ensuring that the data can be cleaned, converted and preprocessed in the most suitable manner for subsequent analysis during the processing process, as well as processed by the prediction model, thereby improving the efficiency of data processing and reducing the transmission delay of data between different processing stages; multiple different types of business volume prediction results are generated based on different data features and model perspectives, and through subsequent fusion processing, they can complement and correct each other, thereby significantly improving the accuracy and reliability of the final prediction results; therefore, the business data processing method of the business processing data has fast business data processing speed and high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A flowchart of the business data processing method described in an embodiment of the present application is shown;

[0040] Figure 2 A schematic diagram showing the structure of the business data processing system according to an embodiment of the present application is shown;

[0041] Figure 3 Another flow chart of the business data processing method described in the embodiment of the present application is shown;

[0042] Figure 4 A flowchart of another method for processing business data according to an embodiment of the present application is shown;

[0043] Figure 5 A schematic diagram showing the structure of the service data processing device according to an embodiment of the present application is shown;

[0044] Figure 6 A schematic structural diagram of an electronic device described in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0045] To make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the drawings in the present application only serve the purpose of explanation and description and are not used to limit the scope of protection of the present application. In addition, it should be understood that the schematic drawings are not drawn in real proportion. The flowchart used in this application shows the operations implemented according to some embodiments of the present application. It should be understood that the operations of the flowchart can be implemented out of sequence, and the steps without logical context can be reversed in order or implemented simultaneously. In addition, those skilled in the art can add one or more other operations to the flowchart under the guidance of the content of the present application, or remove one or more operations from the flowchart.

[0046] In addition, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.

[0047] It should be noted that the term "comprising" will be used in the embodiments of the present application to indicate the existence of the features declared thereafter, but does not exclude the addition of other features.

[0048] As the scale of enterprises expands and market competition intensifies, in the context of the big data era, the amount of data generated by enterprises every day is growing exponentially, covering multiple dimensions such as sales, production, supply chain, and market feedback. Although the existing financial forecasting system has designed some forecasting algorithms to keep up with the times, it is often difficult to quickly process these massive amounts of data due to outdated architecture design and inefficient algorithms. As a result, the existing financial forecasting technology is limited by the slow processing speed and insufficient accuracy of business data, and is unable to provide decision support information in a timely manner. This lag makes it difficult for enterprises to respond quickly and accurately when facing market opportunities or risks.

[0049] Based on this, an embodiment of the present application provides a business data processing method, the method comprising obtaining target business data from multiple different data sources; sending the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model server types; the different types of data processing servers respectively process the target business data based on the configured data processing rules, respectively determine the target business feature data that matches the different types of AI model server types, and send the target business feature data to the matching AI model server; the different types of AI model servers respectively process the received business feature data based on a pre-configured and trained business volume prediction model, Determine different types of business volume prediction results; fuse the different types of business volume prediction results to determine the target business volume prediction results; the method can instantly capture, clean, convert and predict data streams from multiple sources; various types of data processing servers are configured with data processing rules that match specific types of AI model servers, ensuring that the data can be cleaned, converted and preprocessed in the most suitable manner for its subsequent analysis and processed by the prediction model during the processing process, thereby improving the efficiency of data processing and reducing the transmission delay of data between different processing stages; multiple different types of business volume prediction results are generated based on different data features and model perspectives, and through subsequent fusion processing, they can complement and correct each other, thereby significantly improving the accuracy and reliability of the final prediction results; therefore, the business data processing method of the business processing data has fast business data processing speed and high accuracy.

[0050] Please refer to Figure 1 , Figure 1 A flow chart of the business data processing method described in an embodiment of the present application is shown; Figure 1 As shown, the business data processing method described in the embodiment of the present application includes the following steps S101-S105:

[0051] S101, acquiring target business data from multiple different data sources;

[0052] S102, sending the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers;

[0053] S103, the different types of data processing servers process the target business data based on the configured data processing rules, respectively determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type;

[0054] S104, the different types of AI model servers respectively process the received service feature data based on the pre-configured and trained service volume prediction models to determine different types of service volume prediction results;

[0055] S105: Fusion the different types of business volume prediction results to determine a target business volume prediction result.

[0056] The business data processing method described in the embodiment of the present application is applied to the business data processing system, please refer to Figure 2 , Figure 2 The structure diagram of the business data processing system according to the embodiment of the present application is shown; Figure 2 As shown, the business data processing system includes a data collection cluster server, a system database, multiple first data processing servers, at least one second data processing server, a first AI model server, a second AI model server, a model fusion and self-adjustment server, and a user terminal; the data collection cluster server is connected to multiple different data sources, and the data collection cluster server is connected to the system database; the system database is connected to the first data processing server and the second data processing server; the first data processing server is connected to the first AI model server, and the second data processing server is connected to the second AI model server, and the first AI model server and the second AI model server are both connected to the user terminal.

[0057] Based on this, the business data processing system described in the embodiment of the present application adopts a modular design, mainly including a data processing module (data processing server), a prediction model module (AI model server) and a result display module (user terminal). The relevant modules are deployed independently and have elastic scaling capabilities. The computing resources and storage resources can be dynamically adjusted according to changes in business needs and data volume to ensure the high availability and scalability of the system.

[0058] In step S101, target business data is acquired from a plurality of different data sources.

[0059] The different data sources include: various business systems of the company, cooperative third-party data sources, and third-party channels such as market research institutions and industry associations (i.e. Figure 2 other data sources in ).

[0060] Exemplarily, the business data includes a general ledger account table, a insurance type information table, a financial original voucher table, a transaction information table, and also includes a base table set by the system for receiving third-party data.

[0061] The business data includes first-type business data and second-type business data; the first-type business data is sequential business data with time series characteristics, and the second-type business data is non-sequential business data in a specific scenario.

[0062] Specifically, the sequential business data with time series characteristics includes daily business data, monthly business data, quarterly business data, etc. The non-sequential business data in the specific scenario includes business data under holiday effects (such as the impact of holidays such as the Spring Festival and National Day on sales), business data under the influence of marketing activities (promotional activities, new product launches, etc.), etc.

[0063] The target business data is acquired from a plurality of different data sources, that is, the data collection cluster server collects the target business data from different data sources and sends the target business data to the system database, and the system database acquires the target business data.

[0064] Specifically, the data collection cluster server integrates advanced real-time data access technologies, such as API interfaces and message queues, so that the business data processing system can instantly capture and process data streams from multiple sources, ensuring that the prediction results are based on the latest and most accurate information, providing instant feedback for corporate decision-making.

[0065] In step S102, the target business data is sent to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model server types.

[0066] Please refer to Figure 3 , Figure 3 Another flow chart of the business data processing method described in the embodiment of the present application is shown; Figure 3 As shown, after data collection, the data processing module (ie, the first data processing server and the second data processing server) is entered.

[0067] In step S103, the different types of data processing servers process the target business data based on the configured data processing rules, respectively determine the target business feature data that matches the different types of AI model servers, and send the target business feature data to the AI ​​model servers of the matching type.

[0068] For details, please refer to Figure 2The first data processing server processes the target business data to obtain the target business feature data required by the business volume prediction model in the first AI model server; the second data processing server processes the target business data to obtain the target business feature data required by the business volume prediction model in the second AI model server.

[0069] Please refer to Figure 3 Different types of data processing servers process the target business data based on configured data processing rules. Specifically, the processing process includes data cleaning, smoothing, data filtering and feature engineering.

[0070] The data cleaning is to thoroughly clean the target business data (such as the general ledger account table, insurance type information table, financial original voucher table, transaction information table, and also the base table set up by this system for receiving third-party data) to remove duplications, errors and outliers to ensure the purity of the data.

[0071] The specific data format of the cleaned target business data may vary from system to system, but it usually includes the following key fields:

[0072] Timestamp: such as recording the specific time when the sale occurred (year, month, day, hour, minute, etc.);

[0073] Product type: Identifies the type of product sold (such as term life insurance, whole life insurance, annuity insurance, endowment insurance, and universal insurance, etc.);

[0074] Sales volume: the number or units sold during a specified period of time;

[0075] Sales: The total revenue or amount realized from sales during a specified period of time;

[0076] Other relevant information: such as sales channels (online / offline), sales personnel, customer information, etc.

[0077] When smoothing the cleaned data, the seasonal smoothing method is first used. Since economic variables show horizontal changes and are affected by seasonal fluctuations, the problem is solved by processing the time series data, separating the seasonal fluctuation factors from the variable level change process, so that the processed sequence data only reflects the level change process, and then use the exponential smoothing method to predict.

[0078] The smoothing process is specifically shown in the following formula (1):

[0079]

[0080] Among them, T represents the Tth period; X Trepresents the target business data after the smoothing process of the Tth period, L is the length of the seasonal fluctuation cycle (such as the number of months or seasons); I is the seasonal adjustment factor, which can be a seasonal ratio or a seasonal index, I T-L It is the data corresponding to the Tth period that only reflects seasonal fluctuations. The original time series data of the corresponding period is divided by this index, and the result is the time series data that only reflects the horizontalization process; S T-1 The historical target business data of period T-1 before the smoothing process of period T is characterized and integrated into the forecast according to the geometric series of (1-α). For example, the weight of period t is α(1-α)0, that of period t-1 is α(1-α)1, that of period t-2 is α(1-α)2, and so on. Because the value of 1-α is between 0 and 1, the higher the number, the smaller the weight of the demand history, and it decays at a geometric rate.

[0081] After obtaining the target business data after smoothing, filtering is performed. The filtering may adopt a low-pass filter and / or a logarithmic transformation according to the characteristics of the target business data.

[0082] Low-pass filter: allows low-frequency signals to pass through while suppressing high-frequency noise. In the sales volume and sales life insurance data, low-frequency signals may represent long-term trends, while high-frequency noise may be caused by random factors.

[0083] Logarithmic transformation:

[0084] If the sales volume or sales data presents a skewed distribution (such as right skewness), logarithmic transformation can be used to improve its distribution characteristics and make the data closer to normal distribution. This helps the stability of subsequent statistical analysis.

[0085] After obtaining the filtered target business processing, key features are extracted and constructed based on feature engineering.

[0086] Exemplarily, the key features include but are not limited to seasonal trends (such as annual, quarterly, and monthly periodic changes), holiday effects (such as the impact of holidays such as the Spring Festival and National Day on sales), marketing activity effects (promotional activities, new product launches, etc.), etc., which provide rich and effective input for model training.

[0087] In step S104, the different types of AI model servers process the received business feature data based on pre-configured and trained business volume prediction models to determine different types of business volume prediction results.

[0088] In some embodiments, in the business data processing method, the different types of AI model servers process the received business feature data based on pre-configured and trained business volume prediction models to determine different types of business volume prediction results, including:

[0089] Each type of AI model server is pre-configured with multiple trained business volume prediction models, and the multiple business volume prediction models respectively process the received business feature data to determine multiple business volume prediction results;

[0090] The multiple business volume prediction results are integrated to determine the business volume prediction result of this type of AI model server.

[0091] That is to say, the multiple models in each AI model server may interpret the data from different angles or focuses, and integrating these prediction results can form a more comprehensive and accurate business volume forecast.

[0092] Please refer to Figure 2 and Figure 3 Different types of data processing servers correspond to different types of AI model servers. Specifically, the different types of AI model servers include a first AI model server for sequential business data and a second AI model server for non-sequential business data; different types of models are integrated to process the target business data.

[0093] The traffic volume prediction model in the first AI model server includes a first traffic volume prediction model obtained by training based on the LSTM model and a second traffic volume prediction model obtained by training based on the GRU model;

[0094] The traffic volume prediction model in the second AI model server includes a third traffic volume prediction model trained based on a random forest model and a fourth traffic volume prediction model trained based on a linear regression model.

[0095] For the first AI model server, LSTM and GRU are both recurrent neural network (RNN) models used to process sequence data. Therefore, their input features are usually serialized data, that is, the data needs to be arranged in chronological order. Sequential business data has time order or sequence characteristics, that is, there is a temporal relationship between data points. LSTM and GRU can effectively capture the long-term dependencies and temporal characteristics in the data through their internal "gate" mechanisms (such as forget gates, input gates, and output gates), and make accurate predictions for business data with sequence characteristics.

[0096] The sequence business data, for example, stock prices, time series sensor data, user behavior logs, etc.

[0097] Training process: Use business data such as historical sales volume and sales revenue as training sets to train the LSTM and GRU models, adjust model parameters through the back propagation algorithm, optimize prediction performance, and obtain a trained business volume prediction model.

[0098] Real-time prediction: After the model training is completed, it is deployed to the system to receive new sales volume and sales data in real time, make rolling predictions, and capture dynamic changes in the market.

[0099] Based on the LSTM and GRU models, combined with linear regression or random forest models, regression analysis is performed on other important factors that affect premium sales and other business volumes (such as holiday effects, promotional strategies, economic indicators, etc.). The training adjusts the model parameters through Bayesian optimization to optimize the model performance.

[0100] Linear regression and random forest are not specific to sequential data. Non-sequential business data does not have a clear time order, or even if there is a timestamp, the data points do not rely on a temporal relationship for analysis. Models such as linear regression and random forest are used to process this type of data. Non-sequential business data can be numerical, categorical, or mixed, such as sales, number of users, product features, etc.

[0101] In this way, different types of AI model servers process the received business feature data based on pre-configured and trained business volume prediction models to determine different types of business volume prediction results.

[0102] It should be noted that due to the relatively small amount of business data affected by other important factors (such as holiday effects, promotional strategies, economic indicators, etc.), please refer to Figure 2 , the number of first data processing servers corresponding to LSTM and GRU models is greater than the number of second data processing servers corresponding to linear regression and random forest models.

[0103] By modularly deploying the first data processing server and the second data processing server, the business data processing system has efficient real-time data processing capabilities, including steps such as data cleaning, conversion and feature extraction, and can complete complex data processing tasks in a very short time, ensuring the smoothness and efficiency of the prediction process.

[0104] In step S105, the different types of business volume prediction results are integrated to determine a target business volume prediction result.

[0105] The prediction results of the time series analysis model in the first AI model server are integrated with the prediction results of the regression analysis model in the second AI model server. Specifically, this can be achieved through a weighted average method to fully utilize the advantages of the two models.

[0106] The fused model can more comprehensively consider the temporal characteristics and non-temporal factors in the data, thereby improving the accuracy and robustness of the prediction.

[0107] The target business volume forecast result includes key business indicators, as well as information such as forecast interval and confidence level. For example, the key business indicators include sales volume, sales revenue, etc.

[0108] The business data processing method described in the embodiment of the present application adopts a prediction model that has been fully trained with historical data, and combines real-time data processing and feature engineering technology, so that the system can accurately capture market dynamics and output high-precision prediction results, helping enterprises to accurately grasp market trends and sales conditions.

[0109] In some embodiments, the business volume prediction model in the AI ​​model server supports online learning, can receive new sample data in real time and automatically adjust parameters to adapt to market changes.

[0110] Please refer to Figure 4 In some embodiments, in the business data processing method, after determining the target business volume prediction result, the method further includes the following steps S401-S403:

[0111] S401, acquiring actual business data corresponding to the target business volume prediction result from a plurality of different data sources;

[0112] S402: Process the actual business data to generate sample data corresponding to prediction models in different types of AI model servers, respectively, to obtain different types of sample data;

[0113] S403: Training a business volume prediction model based on type-matched sample data obtained in a preset time step, so as to update the business volume prediction model in the AI ​​model server according to the preset time step.

[0114] At each time step, the traffic prediction model only uses the currently received data samples to calculate the gradient and updates the model parameters accordingly. The online gradient descent algorithm only uses a small amount of data samples for update each time, so the calculation amount is small and the update speed is fast.

[0115] In some embodiments, the training of the traffic volume prediction model based on the type-matched sample data obtained in the preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step includes:

[0116] The traffic volume prediction model processes the sample data of the type matching received in the preset time step to determine the loss result of the loss function of the traffic volume prediction model under the current parameters;

[0117] Calculate the gradient of the loss function with respect to the current parameter based on the loss result;

[0118] Based on the gradient, and the pre-configured update rules and learning rate of the gradient descent method, the parameters of the business volume prediction model are updated to update the business volume prediction model in the AI ​​model server according to a preset time step.

[0119] The specific steps of training the traffic volume prediction model based on the type-matched sample data obtained in the preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step are as follows:

[0120] Step 1: Initialize parameters: The parameters θ0 of the traffic prediction model are usually initialized to zero vectors or random values.

[0121] Step 2. Sample data arrives one by one: At each time step t, a new sample data (x t ,y t ), where x t is the eigenvector, y t is the corresponding label or true value.

[0122] Step 3. Calculate the loss function: Based on the current parameters θ of the traffic forecasting model t Calculate the loss function to determine the prediction error of the model;

[0123] Exemplarily, the loss function may be a square loss, specifically the following formula (2):

[0124]

[0125] Step 4. Calculate the gradient: for the loss function L t Calculate the current parameter θ t Gradient gradient Indicates the direction in which the loss changes with the parameters.

[0126] Step 5. Update parameters: According to the update rule of the gradient descent method, use the learning rate η to update the parameters, specifically the following formula (3):

[0127]

[0128] Among them, the learning rate η controls the step size. If it is too large, it will easily lead to oscillation, and if it is too small, the convergence will be too slow.

[0129] Step 6. Repeat steps: Every time new data arrives, repeat the above steps and continuously update the parameters.

[0130] The business data processing system described in the embodiment of the present application supports online learning algorithms. The model deployed in the AI ​​model server can automatically adjust parameters as new data is continuously input, adapt to market changes, maintain the accuracy and timeliness of predictions, and eliminate the need for regular manual updates of the model, thereby improving the system's degree of automation and flexibility.

[0131] In some embodiments, in the business data processing method, after acquiring actual business data corresponding to the target business volume prediction result from a plurality of different data sources, the method further includes:

[0132] Comparison results of comparing the actual traffic volume in the actual traffic data with traffic volume prediction results of traffic volume prediction models in different types of AI model servers;

[0133] Based on the comparison result, a feedback result for the traffic volume prediction model in the AI ​​model server is generated.

[0134] That is to say, the business data processing method described in the embodiment of the present application establishes a feedback mechanism, analyzes the prediction error through the comparison mechanism of the business volume prediction results and the actual data, and uses the feedback results for continuous optimization and adjustment of the model.

[0135] In some embodiments, in the business data processing method, after determining the target business volume prediction result, the method further includes:

[0136] The target traffic volume prediction result is processed into a chart and / or a report, and the chart and / or the report is sent to a terminal device so as to display the target traffic volume prediction result on a graphical user interface of the terminal device.

[0137] In other words, the forecast results are displayed in real time on the user interface in the form of charts, reports, etc., so that users can quickly understand market trends and sales situations.

[0138] The business data processing method described in the embodiment of the present application converts complex forecast data into charts, reports, etc. through an intuitive and easy-to-understand forecast result display interface provided by the business data processing system, so as to facilitate corporate decision makers to quickly understand market trends and sales conditions, and provide strong support for formulating scientific and reasonable business strategies.

[0139] It should be noted that, in some embodiments, during the data transmission, storage and processing, the business data processing method adopts encryption technology and access control strategy to ensure the security and privacy of data, prevent the leakage of sensitive information, and protect the business interests of the enterprise.

[0140] Based on the same inventive concept, a business data processing device corresponding to the business data processing method is also provided in the embodiment of the present application. Since the principle of solving the problem by the device in the embodiment of the present application is similar to the above-mentioned business data processing method in the embodiment of the present application, the implementation of the device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0141] Please refer to Figure 5 , Figure 5 A schematic diagram of the structure of the service data processing device according to an embodiment of the present application is shown; Figure 5 As shown, the business data processing device includes:

[0142] An acquisition module 501 is used to acquire target business data from multiple different data sources;

[0143] A sending module 502 is used to send the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers;

[0144] The first processing module 503 is used to enable the different types of data processing servers to process the target business data based on the configured data processing rules, respectively determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type;

[0145] A determination module 504 is used to enable the different types of AI model servers to process the received service feature data based on the pre-configured and trained service volume prediction models to determine different types of service volume prediction results;

[0146] The fusion module 505 is used to fuse the different types of business volume prediction results to determine a target business volume prediction result.

[0147] In some embodiments, the business data processing device further includes:

[0148] An updating module, configured to obtain actual business data corresponding to the target business volume prediction result from a plurality of different data sources after determining the target business volume prediction result;

[0149] Processing the actual business data, generating sample data corresponding to prediction models in different types of AI model servers, and obtaining different types of sample data;

[0150] The traffic volume prediction model is trained based on type-matched sample data obtained in a preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step.

[0151] In some embodiments, the update module in the business data processing device is specifically used to train the business volume prediction model based on the type-matched sample data obtained in the preset time step to update the business volume prediction model in the AI ​​model server according to the preset time step:

[0152] The traffic volume prediction model processes the sample data of the type matching received in the preset time step to determine the loss result of the loss function of the traffic volume prediction model under the current parameters;

[0153] Calculate the gradient of the loss function with respect to the current parameter based on the loss result;

[0154] Based on the gradient, and the pre-configured update rules and learning rate of the gradient descent method, the parameters of the business volume prediction model are updated to update the business volume prediction model in the AI ​​model server according to a preset time step.

[0155] In some embodiments, the business data processing device further includes a comparison module, which is used to obtain actual business data corresponding to the target business volume prediction result from multiple different data sources, and then compare the actual business volume in the actual business data with the business volume prediction results of the business volume prediction models in different types of AI model servers;

[0156] Based on the comparison result, a feedback result for the traffic volume prediction model in the AI ​​model server is generated.

[0157] In some embodiments, the determination module in the business data processing device is specifically used to:

[0158] Each type of AI model server is pre-configured with multiple trained business volume prediction models, and the multiple business volume prediction models respectively process the received business feature data to determine multiple business volume prediction results;

[0159] The multiple business volume prediction results are integrated to determine the business volume prediction result of this type of AI model server.

[0160] In some embodiments, in the business data processing device, the different types of AI model servers include a first AI model server for sequential business data and a second AI model server for non-sequential business data;

[0161] The traffic volume prediction model in the first AI model server includes a first traffic volume prediction model obtained by training based on the LSTM model and a second traffic volume prediction model obtained by training based on the GRU model;

[0162] The traffic volume prediction model in the second AI model server includes a third traffic volume prediction model obtained by training based on a random forest model and a fourth traffic volume prediction model obtained by training based on a linear regression model.

[0163] In some embodiments, the business data processing device further includes:

[0164] The second processing module is used to process the target business volume forecast result into a chart and / or report after determining the target business volume forecast result, and send the chart and / or report to the terminal device to display the target business volume forecast result on the graphical user interface of the terminal device.

[0165] Based on the same inventive concept, an electronic device corresponding to the business data processing method is also provided in the embodiment of the present application. Since the principle of solving the problem by the electronic device in the embodiment of the present application is similar to the above-mentioned business data processing method in the embodiment of the present application, the implementation of the electronic device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0166] Please refer to Figure 6 , Figure 6 A schematic diagram of the structure of an electronic device described in an embodiment of the present application is shown; the electronic device 600 includes: a processor 602, a memory 601 and a bus, the memory 601 stores machine-readable instructions executable by the processor 602, and when the electronic device 600 is running, the processor 602 communicates with the memory 601 through the bus, and when the machine-readable instructions are executed by the processor 602, the steps of the business data processing method are executed.

[0167] Based on the same inventive concept, a computer-readable storage medium corresponding to the business data processing method is also provided in the embodiment of the present application. Since the principle of solving the problem by the computer-readable storage medium in the embodiment of the present application is similar to the above-mentioned business data processing method in the embodiment of the present application, the implementation of the computer-readable storage medium can refer to the implementation of the method, and the repeated parts will not be repeated.

[0168] A computer-readable storage medium stores a computer program, which executes the steps of the business data processing method when executed by a processor.

[0169] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working process of the system and device described above can refer to the corresponding process in the method embodiment, and will not be repeated in this application. In the several embodiments provided in this application, it should be understood that the disclosed system, device and method can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation. For example, multiple modules 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 communication interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0170] The modules described as separate components may or may not be physically separated, and the components shown as modules 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.

[0171] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0172] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a platform server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard drives, ROM, RAM, magnetic disks, or optical disks.

[0173] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A business data processing method, characterized in that: The method comprises: Acquire target business data from multiple different data sources; Sending the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers; The different types of data processing servers process the target business data based on the configured data processing rules, respectively, determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type; The different types of AI model servers process the received service feature data based on pre-configured and trained service volume prediction models to determine different types of service volume prediction results; The different types of business volume forecast results are integrated to determine a target business volume forecast result.

2. The business data processing method according to claim 1, characterized in that: After determining the target business volume prediction result, the method further includes: Acquire actual business data corresponding to the target business volume prediction result from a plurality of different data sources; Processing the actual business data, generating sample data corresponding to prediction models in different types of AI model servers, and obtaining different types of sample data; The traffic volume prediction model is trained based on type-matched sample data obtained in a preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step.

3. The business data processing method according to claim 1, characterized in that: The training of the traffic volume prediction model based on the sample data of the type matching obtained in the preset time step to update the traffic volume prediction model in the AI ​​model server according to the preset time step includes: The traffic volume prediction model processes the sample data of the type matching received in the preset time step to determine the loss result of the loss function of the traffic volume prediction model under the current parameters; Calculate the gradient of the loss function with respect to the current parameter based on the loss result; Based on the gradient, and the pre-configured update rules and learning rate of the gradient descent method, the parameters of the business volume prediction model are updated to update the business volume prediction model in the AI ​​model server according to a preset time step.

4. The business data processing method according to claim 2, characterized in that: After acquiring actual business data corresponding to the target business volume prediction result from a plurality of different data sources, the method further includes: Comparison results of comparing the actual traffic volume in the actual traffic data with traffic volume prediction results of traffic volume prediction models in different types of AI model servers; Based on the comparison result, a feedback result for the traffic volume prediction model in the AI ​​model server is generated.

5. The business data processing method according to claim 1, characterized in that: The different types of AI model servers process the received service feature data based on pre-configured and trained service volume prediction models to determine different types of service volume prediction results, including: Each type of AI model server is pre-configured with multiple trained business volume prediction models, and the multiple business volume prediction models respectively process the received business feature data to determine multiple business volume prediction results; The multiple business volume prediction results are integrated to determine the business volume prediction result of this type of AI model server.

6. The business data processing method according to claim 5, characterized in that: The different types of AI model servers include a first AI model server for sequential business data and a second AI model server for non-sequential business data; The traffic volume prediction model in the first AI model server includes a first traffic volume prediction model obtained by training based on the LSTM model and a second traffic volume prediction model obtained by training based on the GRU model; The traffic volume prediction model in the second AI model server includes a third traffic volume prediction model trained based on a random forest model and a fourth traffic volume prediction model trained based on a linear regression model.

7. The business data processing method according to claim 1, characterized in that: After determining the target business volume prediction result, the method further includes: The target traffic volume prediction result is processed into a chart and / or a report, and the chart and / or the report is sent to a terminal device so as to display the target traffic volume prediction result on a graphical user interface of the terminal device.

8. A business data processing device, characterized in that: The device comprises: An acquisition module is used to acquire target business data from multiple different data sources; A sending module, used to send the target business data to multiple types of data processing servers; wherein different types of data processing servers correspond to different types of AI model servers, and different types of data processing servers are configured with data processing rules that match different types of AI model servers; A first processing module is used to enable the different types of data processing servers to process the target business data based on the configured data processing rules, respectively, determine the target business feature data that matches the type of the different types of AI model servers, and send the target business feature data to the AI ​​model servers that match the type; A determination module, configured to enable the different types of AI model servers to process the received service feature data based on pre-configured and trained service volume prediction models, respectively, to determine different types of service volume prediction results; The fusion module is used to fuse the different types of business volume prediction results to determine the target business volume prediction result.

9. An electronic device, characterized in that: include: A processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor and the memory communicate via the bus, and when the machine-readable instructions are executed by the processor, the steps of the business data processing method as described in any one of claims 1 to 7 are performed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the business data processing method as described in any one of claims 1 to 7 are executed.

Images