Business prediction method and device, electronic equipment, medium and product

Abstract

The invention provides a service prediction method and device, electronic equipment, a medium and a product. The method comprises the following steps: determining a historical business trend chart of a target business according to historical index data of a business index of the target business in a historical time period; according to the historical business trend chart and a prediction time period, generating a prediction business trend chart of the target business; and predicting the business trend of the target business according to the predicted business trend chart to obtain a business prediction result of the target business. According to the embodiment of the invention, the business prediction efficiency and accuracy can be improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and in particular to a business forecasting method and device, an electronic device, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the rapid development and popularization of various Internet service businesses (such as finance, commodity sales, content distribution / subscription, etc.), time series analysis has gradually become an important means to predict users' business needs. Among them, time series prediction is the most common and important application scenario in time series analysis tasks. By mining the potential laws of historical behaviors related to users' business needs, users' business needs can be predicted and judged, thereby improving user service levels.

[0003] However, there are various potential connections between the variables involved in time series data, which often exhibit one or more characteristics, such as massiveness, trend, periodicity, volatility, stability and symmetry, making prediction difficult. The prediction algorithms in related technologies can hardly meet the prediction needs. Summary of the invention

[0004] The present disclosure provides a business forecasting method and device, an electronic device, a computer-readable storage medium, and a computer program product.

[0005] In a first aspect, the present disclosure provides a business forecasting method, the method comprising:

[0006] Determine a historical business trend graph of the target business based on historical indicator data of the target business within a historical time period;

[0007] Generate a forecasted business trend graph of the target business according to the historical business trend graph and the forecast time period;

[0008] According to the predicted business trend graph, the business trend of the target business is predicted to obtain a business prediction result of the target business.

[0009] In a second aspect, the present disclosure provides a business prediction device, the device comprising:

[0010] A historical trend determination module, used to determine a historical business trend graph of the target business based on the historical indicator data of the business indicator of the target business within a historical time period;

[0011] A forecast trend generation module, used to generate a forecast business trend graph of the target business based on the historical business trend graph and the forecast time period;

[0012] The result determination module is used to predict the business trend of the target business according to the predicted business trend graph to obtain the business prediction result of the target business.

[0013] In a third aspect, the present disclosure provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores one or more computer programs executable by the at least one processor, and one or more of the computer programs are executed by the at least one processor so that the at least one processor can execute the above-mentioned business forecasting method.

[0014] In a fourth aspect, the present disclosure provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the above-mentioned business prediction method when executed by a processor.

[0015] In a fifth aspect, the present disclosure provides a computer program or a computer program product, wherein the computer program product comprises a computer program, the computer program is stored in a computer-readable storage medium, and the computer program implements the above-mentioned business forecasting method when executed by a processor.

[0016] The embodiments provided by the present disclosure can determine a historical business trend graph based on historical indicator data of the target business, generate a predicted business trend graph based on the historical trend graph and the predicted time period, and realize time series prediction through image generation, thereby fully capturing the correlation between time series data; and, performing business trend prediction based on the predicted business trend graph can improve the efficiency and accuracy of business prediction.

[0017] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The accompanying drawings are used to provide a further understanding of the present disclosure and constitute a part of the specification. Together with the embodiments of the present disclosure, they are used to explain the present disclosure and do not constitute a limitation of the present disclosure. The above and other features and advantages will become more apparent to those skilled in the art by describing detailed example embodiments with reference to the accompanying drawings, in which:

[0019] Figure 1 The application scenario diagram of the business prediction method and device provided by the embodiment of the present disclosure is schematically shown;

[0020] Figure 2 A flowchart of a business prediction method provided by an embodiment of the present disclosure;

[0021] Figure 3a A schematic diagram of a historical business trend diagram of a business forecasting method provided by an embodiment of the present disclosure;

[0022] Figure 3b A schematic diagram of a predicted business trend graph of a business prediction method provided in an embodiment of the present disclosure;

[0023] Figure 4a , Figure 4b and Figure 4c A schematic diagram of a mask for a business prediction method provided by an embodiment of the present disclosure;

[0024] Figure 5 A block diagram of a business prediction device provided by an embodiment of the present disclosure;

[0025] Figure 6 A block diagram of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the exemplary embodiments of the present disclosure are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding, which should be considered as merely exemplary. Therefore, those of ordinary skill in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, the description of well-known functions and structures is omitted in the following description.

[0027] In the absence of conflict, the various embodiments of the present disclosure and the various features therein may be combined with each other.

[0028] As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0029] The terms used herein are only used to describe specific embodiments and are not intended to limit the present disclosure. As used herein, the singular forms "a" and "the" are also intended to include plural forms, unless the context clearly indicates otherwise. It will also be understood that when the terms "including" and / or "made of" are used in this specification, the presence of the features, wholes, steps, operations, elements and / or components is specified, but the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or groups thereof is not excluded. "Connected" or "connected" and similar words are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0030] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by those of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant art and the present disclosure, and will not be interpreted as having an idealized or overly formal meaning unless explicitly defined as such herein.

[0031] In the technical solution disclosed in this disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved are in compliance with the provisions of relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution complies with relevant national laws and regulations (for example, the "Information Security Technology Personal Information Security Specification", etc.). For example: corresponding prescribed measures are taken for personal information access control; prescribed restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; when using personal information, clear identity orientation is eliminated to avoid precise positioning of specific individuals.

[0032] There are inherent potential connections between the variables involved in time series data, which often exhibit one or more characteristics, such as massiveness, trend, periodicity, volatility, stability, and symmetry. Based on this, in related technologies, a variety of traditional algorithms and deep learning algorithms have been proposed to solve time series prediction problems.

[0033] At present, the traditional algorithms commonly used in related technologies include moving average method, Holt linear trend method, seasonal exponential average method, ARIMA model (Autoregressive Integrated Moving Average Model), etc. These methods are difficult to meet the requirements of high efficiency and high accuracy of prediction tasks when processing nonlinear, high-noise multidimensional data.

[0034] Deep learning algorithms can effectively solve the problems existing in traditional algorithms. Deep learning algorithms used for time series prediction can be divided into three categories:

[0035] 1) Convolutional Neural Network (CNN) is a type of feedforward neural network with convolution and pooling operations as its core. It shows ideal results when dealing with short-term prediction tasks, but it is difficult to handle long-step time series prediction problems;

[0036] 2) Recurrent Neural Network (RNN) is a deep recursive model that processes variable-length sequence data. As the main model for time series prediction, RNN has shown excellent performance in long-term and short-term prediction tasks, but high memory consumption and gradient transfer problems limit its further application;

[0037] 3) Transformer (Transformer model) is a type of sequence model based on the attention mechanism, which has a good ability to capture long-term and short-term dependencies, but the algorithm has high computational complexity and has high requirements on the equipment.

[0038] For time series prediction tasks, related technologies tend to start from data at a time point and use sequence data as model input for prediction. The data at each time point will have a greater impact on the prediction results. Therefore, the sequence model is more sensitive to the lack of data samples. In addition, the prediction time step and the input time series length will affect the final performance of the model, which may lead to poor prediction results.

[0039] According to the embodiments of the present disclosure, it is possible to realize time series prediction by computer image processing from the perspective of image processing. During the processing, the historical indicator data of the target business is plotted into a historical business trend graph; based on the historical trend graph and the predicted time period, the predicted business trend graph is generated by the image generation model, thereby fully capturing the specificity and relevance of different historical behaviors of users, and is not easily affected by the length of the time series. Moreover, by adjusting the size of the predicted time period, the prediction of data with different time steps can be realized, thus overcoming the problem of inaccurate prediction of long step lengths in the time series prediction model in the related technology.

[0040] Furthermore, according to the embodiments of the present disclosure, it is possible to combine the advantages of a multimodal big model and utilize the multimodal big model to predict business trend graphs and business needs for specific tasks (such as completing corresponding result analysis), thereby achieving the output of business prediction results for the target business, thereby improving business prediction efficiency and prediction accuracy.

[0041] Figure 1 The application scenario diagram of the business prediction method and device provided by the embodiment of the present disclosure is schematically shown.

[0042] like Figure 1 As shown, the application scenario of the embodiment of the present disclosure may include a terminal device 101, a network 103 and a server 102. The network 103 is used to provide a medium for a communication link between the terminal device 101 and the server 102. The network 103 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0043] The user can use the terminal device 101 to interact with the server 102 through the network 103 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only as examples).

[0044] The terminal device 101 may be any electronic device having a display screen and supporting web browsing, including but not limited to a smart phone, a tablet computer, a laptop computer, a desktop computer, and the like.

[0045] The server 102 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the terminal device 101. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0046] It should be noted that the business prediction method and device provided in the embodiment of the present disclosure can be executed by the server 102. Accordingly, the business prediction device provided in the embodiment of the present disclosure can be set in the server 102. The business prediction method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 102 and can communicate with the terminal device 101 and / or the server 102. Accordingly, the business prediction device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 102 and can communicate with the terminal device 101 and / or the server 102.

[0047] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0048] Figure 2 A flowchart of a business prediction method provided by an embodiment of the present disclosure. Figure 2 , the method comprising:

[0049] Step S21, determining a historical business trend graph of the target business according to the historical indicator data of the business indicator of the target business within a historical time period;

[0050] Step S22, generating a predicted business trend graph of the target business according to the historical business trend graph and the predicted time period;

[0051] Step S23: predicting the business trend of the target business according to the predicted business trend graph to obtain a business prediction result of the target business.

[0052] For example, the target business to be analyzed may include financial business, and may also include commodity sales business, content distribution / subscription business, system operation and maintenance business, etc. The present disclosure does not limit the specific type of the target business.

[0053] In some possible implementations, the target business to be analyzed may have various corresponding business indicators. Through the business system or business platform of the target business, various business data of a large number of users when using or handling the target business within a historical time period (e.g., 6 months, 1 year, 2 years, etc.) can be collected, and the cumulative amount of each business indicator in the statistical unit can be counted in a preset statistical unit (e.g., 1 day, 1 week, 1 month, etc.), so as to obtain historical indicator data represented in the form of a time series. In the example, when the historical time period is 2 years and the statistical unit is 1 week, the historical indicator data of a certain business indicator can be represented as a time series {a 1 、a 2 ,……,a 104 The present disclosure does not limit the specific duration of the historical time period and the statistical unit.

[0054] In some possible implementations, when the target business is a financial business, the business indicators may include at least one of the number of queries on the object's credit information, the granted quota information of the business resources, the balance information of the business resources, the return amount information of the business resources, and the consumption amount information of the business resources.

[0055] In some possible implementations, the object credit information may be information obtained based on the resource grant, resource return, overdue return, etc. of the object in the relevant business, or credit information fed back by a third-party system. When the object handles the target business, the object credit information of the object will be queried. Accordingly, the business indicator-number of queries on the object credit information may include the total number of queries on the object credit information for all objects handling the target business in each statistical unit.

[0056] In some possible implementations, the amount of business resources granted may be the amount of business resources granted to an object when the object handles the target business, and the business system determines the amount of business resources granted to the object according to its object credit information. Accordingly, the business indicator-business resource grant amount information may include the total amount or statistical average of the business resource grant amounts of all objects handling the target business in each statistical unit. The present disclosure does not limit the specific calculation method of the grant amount information.

[0057] In some possible implementations, the balance of business resources may be the remaining amount that can be applied after the object has applied for a certain amount of business resources within the granted amount of business resources, or the amount of resources that can be applied. Accordingly, the balance information of business indicators-business resources may include the total amount or statistical average of the business resource balances of all objects that have handled the target business in each statistical unit. The present disclosure does not limit the specific calculation method of the balance information.

[0058] In some possible implementations, the amount of service resources returned may be the amount of resources that have been returned by the object that has handled the target service within the agreed time. Accordingly, the service indicator-service resource return amount information may include the total amount or statistical average of the amount of resources returned by all objects that have handled the target service in each statistical unit. The present disclosure does not limit the specific calculation method of the return amount information.

[0059] In some possible implementations, the consumption of business resources may be the amount of resources consumed by an object that has handled the target business after applying for a business resource quota. Accordingly, the business indicator-business resource consumption information may include the total amount or statistical average of the amount of resources consumed by all objects that have handled the target business in each statistical unit. The present disclosure does not limit the specific calculation method of the consumption information.

[0060] It should be understood that the above are only examples of business indicators. Those skilled in the art may set the type and quantity of business indicators and the specific content and calculation method of each business indicator according to actual business conditions, and the present disclosure does not limit this.

[0061] After obtaining historical indicator data of at least one business indicator of the target business within a historical time period, in step S21, a historical business trend graph of the target business may be determined according to the historical indicator data.

[0062] In some possible implementations, different standardization algorithms can be designed to preprocess the historical indicator data of each business indicator so that the standardized indicator data of different business indicators are in different numerical ranges. For example, the standardized indicator data of business indicator 1 is in the numerical range [0, 0.2]; the standardized indicator data of business indicator 2 is in the numerical range [0.2, 0.4], etc. In this way, different business indicators can be prevented from interfering with each other when drawing images, thereby improving the subsequent processing effect.

[0063] In some possible implementations, a historical business trend graph of the target business may be drawn based on the standardized indicator data of each business indicator, and the graph includes a curve corresponding to each business indicator.

[0064] Figure 3aA schematic diagram of a historical business trend graph of a business forecasting method provided in an embodiment of the present disclosure. Figure 3a The horizontal axis represents the number of statistical units in the historical time period, such as the number of weeks; the vertical axis represents the numerical range corresponding to the standardized indicator data. Figure 3a As shown, the figure includes curves corresponding to 5 business indicators.

[0065] In step S22, a forecasted business trend graph of the target business may be generated based on the historical business trend graph and the forecasted time period. The forecasted time period may be a time period of the business trend to be forecasted that is determined based on business needs, for example Figure 3a The historical time period corresponds to weeks 0 to 92, and the forecast time period corresponds to 12 weeks after the historical time period, that is, weeks 93 to 104. The present disclosure does not limit the specific value of the forecast time period.

[0066] In some possible implementations, a candidate trend graph can be obtained based on a historical business trend graph and a predicted time period, wherein the graph includes an image area of ​​an existing curve corresponding to the historical time period and a blank image area corresponding to the predicted time period; and the candidate trend graph is completed by image generation according to an image generation model to obtain a predicted business trend graph of the target business. The image generation model may be a flow-based generative model, a generative adversarial network (GAN), a diffusion model, etc., and the present disclosure does not limit the specific type of the image generation model.

[0067] In step S23, the business trend of the target business is predicted according to the predicted business trend graph to obtain the business prediction result of the target business. That is to say, after obtaining the completed predicted business trend graph, the predicted business trend graph can be further predicted according to specific business needs (such as the application trend of business resources, the overall consumption trend, etc.) to obtain the final business prediction result. Among them, the demand description information and the predicted business trend graph can be simultaneously input into the multimodal big model for processing to obtain the business prediction result; the business prediction result can also be determined by other means. The multimodal big model can be any language-based big model based on GPT, Bert, etc., and the present disclosure does not limit the specific type of the multimodal big model.

[0068] According to the embodiments of the present disclosure, it is possible to determine a historical business trend graph based on historical indicator data of the target business, generate a predicted business trend graph based on the historical trend graph and the predicted time period, and implement time series prediction through image generation, thereby fully capturing the correlation between time series data; and, performing business trend prediction based on the predicted business trend graph can improve the efficiency and accuracy of business prediction.

[0069] The following is a detailed description of the business prediction method according to an embodiment of the present disclosure.

[0070] As mentioned above, historical indicator data of the business indicators of the target business to be analyzed within a historical time period may be obtained, and in step S21, a historical business trend graph of the target business may be determined based on the historical indicator data.

[0071] In some possible implementations, step S21 may include:

[0072] Standardizing the historical indicator data of each business indicator to obtain standardized indicator data of the business indicator; wherein the standardized indicator data of different business indicators are in different value ranges;

[0073] A historical business trend graph is drawn based on the standardized indicator data of the business indicator; wherein the historical business trend graph includes a curve of indicator data changes of each business indicator within the historical time period.

[0074] For example, for each business indicator, the historical indicator data of the business indicator can be standardized according to a preset standardization algorithm to obtain standardized indicator data of the business indicator. The standardization algorithms for different business indicators can be different, and the standardized indicator data of different business indicators are in different numerical ranges. For example, the standardized indicator data of business indicator 1 is in the numerical range [0, 0.2]; the standardized indicator data of business indicator 2 is in the numerical range [0.2, 0.4], etc. In this way, it is possible to avoid interference between different business indicators when drawing an image, thereby improving subsequent processing effects. The present disclosure does not limit the specific type of the standardization algorithm.

[0075] In some possible implementations, a historical business trend graph of the target business can be drawn based on the standardized indicator data of each business indicator. If the data of a certain time point (segment) in the historical indicator data is missing, the missing data is filled by curve fitting or image fitting to ensure the consistency of the sample data distribution. The historical business trend graph can be Figure 3a The line chart shown includes a curve showing the change in indicator data of each business indicator within a historical time period.

[0076] It should be understood that the historical business trend graph may also be in the form of a bar graph, etc. Those skilled in the art may set the specific form of the trend graph according to actual conditions, and the present disclosure does not impose any limitation on this.

[0077] In this way, time series data can be converted into images so that subsequent predictions can be made through image generation, thereby improving the accuracy of the predictions.

[0078] In some possible implementations, after obtaining the historical business trend graph, the trend graph can be completed in step S22 by using an image generation model in the form of image generation. The image generation model can be a flow-based generative model, a generative adversarial network (GAN), a diffusion model, etc. The present disclosure does not limit the specific type of the image generation model.

[0079] Before step S22, the image generation model may be trained in advance. The business prediction method according to the embodiment of the present disclosure may also include: performing unsupervised training on the image generation model according to the historical business trend graph of the target business and the noise sequence to obtain a trained image generation model; wherein the image generation model includes a diffusion model.

[0080] For example, the image generation model may adopt a diffusion model, which includes two processes: forward diffusion (noise addition) and reverse generation (noise removal). That is, in the forward process, the noise or score in the diffusion process is fitted by training the model, thereby achieving denoised image generation in the reverse process.

[0081] Therefore, when the image generation model adopts a diffusion model, a noise sequence can be preset, such as a geometric noise sequence Among them, σ min =σ 1 <σ 2 <…<σ L =σ max , L is the total amount of noise, σ min , σ max are the minimum and maximum values ​​of the noise term, respectively. mib , σ max The specific value of L can be determined according to the actual situation, for example, min =0,σ max =1, L=100, and the present disclosure does not limit its specific value.

[0082] In some possible implementations, the image generation model is unsupervisedly trained based on the historical business trend graph and the noise sequence of the target business. The image generation model may be trained directly based on the historical business trend graph obtained in step S21, or may be trained based on a historical business trend graph drawn based on other historical indicator data of the target business, and the present disclosure does not limit this.

[0083] In some possible implementations, during training, the input training sample may be defined as x, and the data distribution may be defined as p data(x), then the perturbation kernel of the model That is, the perturbation kernel satisfies the normal distribution. The core of the diffusion model is to train the score network model s θ (x,σ), the network model can be obtained by denoising score matching. The optimization objectives of model training are as follows:

[0084]

[0085] In formula (1), x represents the training sample, i.e., the historical business trend graph; represents the noisy training sample after adding the noise term; Denotes the noise term σ i The corresponding data distribution of the noisy training samples; p data (x) represents the data distribution of training samples; Represents the data expectation of the training samples; represents the data expectation of the noisy training sample; N = L represents the number of noise items in the noise sequence; θ* represents the target value of the network parameter of the score network. Formula (1) is used to characterize that the training goal of the model is to minimize the statistical average value of the model's loss function on the noise sequence.

[0086] In some possible implementations, after multiple rounds of unsupervised iterative training until the model converges, the optimal score network model can be obtained as the trained image generation model, expressed as

[0087] It should be understood that those skilled in the art may set the specific training method and optimization target of the model training according to actual conditions, and the present disclosure does not limit this.

[0088] In this way, the training process of the image generation model can be realized, and the accuracy of images generated by subsequent image generation models can be improved, thereby improving the accuracy of business predictions.

[0089] In some possible implementations, after obtaining the trained image generation model, a predicted business trend graph of the target business may be generated in step S22. Step S22 may include:

[0090] Generate a candidate trend graph based on the historical business trend graph and the forecast time period;

[0091] According to the noise sequence, the candidate trend graph is subjected to denoising and generation processing to obtain a predicted business trend graph of the target business.

[0092] For example, the forecast time period may be a time period of business trends to be forecasted that is determined based on business needs, such as Figure 3aThe historical time period corresponds to weeks 0 to 92, and the forecast time period corresponds to 12 weeks after the historical time period, that is, weeks 93 to 104. The present disclosure does not limit the specific value of the forecast time period.

[0093] In some possible implementations, a candidate trend graph may be obtained based on the historical business trend graph and the predicted time period, and the graph includes an image area corresponding to the historical time period and a mask image area corresponding to the predicted time period. The image area corresponding to the historical time period is the image area of ​​the existing curve in the historical business trend graph; the mask image area corresponds to the blank image area to be predicted.

[0094] The mask image area may be set only in the candidate trend graph, or a separate mask image may be set. Figure 4a , Figure 4b and Figure 4c A schematic diagram of a mask of a business prediction method provided by an embodiment of the present disclosure. Figure 4a , Figure 4b and Figure 4c In the mask image shown, a mask m corresponding to the size of the area to be predicted can be set. In the image, the white area consists of 1 and the black area consists of 0. In this way, in the process of generating the image, the white area in the image can be retained by performing element-wise multiplication with the generated image. The multiplication process can be expressed as x a =x b ⊙m,x b is the initial generated image; x a is the generated image after multiplication with the mask m, and ⊙ represents element-wise multiplication.

[0095] In this way, it is possible to ensure that the existing historical business trend graph remains unchanged during the reasoning process, and the model only generates the trend graph for the part related to the prediction time period, avoiding the generated image from deviating from the historical business trend, thereby improving the accuracy of the generated image.

[0096] like Figure 4a , Figure 4b and Figure 4c As shown, the mask image area in the mask image may be the entire image area or a portion of the image area corresponding to the prediction time period, indicating that all or part of the business indicators may be predicted. The present disclosure does not limit the specific setting method of the mask image area.

[0097] In some possible implementations, the candidate trend graph can be denoised and generated based on the noise sequence and the trained image generation model to obtain a predicted business trend graph of the target business, which includes a first image area corresponding to the historical time period and a second image area corresponding to the predicted time period.

[0098] In some possible implementations, the noise sequence includes L noise items, for example, a geometric noise sequence Among them, σ min =σ 1 <σ 2 <…<σ L =σ max , L is the total amount of noise, σ min , σ max are the minimum and maximum values ​​of the noise term, respectively. min , σ max The specific value of L can be determined according to the actual situation, for example, min =0,σ max =1, L=100, and the present disclosure does not limit its specific value.

[0099] In some possible implementations, a predictor-corrector (PC) sampling algorithm may be used to perform denoising and generating processing on the candidate trend graph. The denoising and generating processing includes L rounds of iterations, one round of iteration corresponds to one noise item, and the candidate trend graph includes a mask image area corresponding to the prediction time period. The steps of performing denoising and generating processing on the candidate trend graph according to the noise sequence to obtain the predicted business trend graph of the target business include:

[0100] De-noising the generated image of the previous round of iteration to obtain an intermediate generated image of the current round of iteration; wherein the generated image of the previous round of iteration corresponding to the first round of iteration is the noisy image of the candidate trend graph;

[0101] Determining a noisy image for this round of iteration according to the noise item corresponding to this round of iteration and the candidate trend graph;

[0102] According to the noisy image of this round of iteration and the mask image area, denoising, reconstructing and consistency correction are performed on the intermediate generated image of this round of iteration to obtain the generated image of this round of iteration;

[0103] When the current iteration is the Lth iteration, the generated image of the current iteration is used as the predicted business trend graph.

[0104] For example, before L rounds of iterations, you can prepare the noise sequence used in model training Satisfy σ min =σ 1 <σ 2 <…<σ L =σ max ; σ 0 = 0; set the reference step size ∈; and the noise term σ i-1 The corresponding step size α i-1 Expressed as Set the number of corrections T in each iteration, that is, the number of annealing Langevin dynamics sampling.

[0105] In some possible implementations, the initialization may satisfy the Gaussian distribution (normal distribution) Model input , that is, superimpose the noise term σ on the candidate trend graph max Corresponding noise, get the noisy image of the candidate trend graph So that it can be input into the image generation model for processing in the first round of L iterations.

[0106] In some possible implementations, in each of the L rounds of iterations, a prediction operation and a correction operation are performed. In the prediction operation, the generated image of the previous round of iteration can be denoised by the image generation model to obtain an intermediate generated image of the current round of iteration. Among them, the generated image of the previous round of iteration corresponding to the first round of iteration is the noisy image of the candidate trend map.

[0107] In some possible implementations, the step of denoising the generated image of the previous round of iteration may include:

[0108] By adopting a preset operation method, the generated image of the previous round of iteration is denoised through the noise term of the previous round of iteration and the noise term of the current round of iteration in the noise sequence, so as to obtain the intermediate generated image of the current round of iteration.

[0109] For example, the preset operation method may adopt a numerical stochastic differential equation solving method, and the present disclosure does not limit the specific operation method.

[0110] For any round of iteration (i-1th round, 1≤i≤L), the prediction operation can be performed by using numerical SDE solvers to solve the generated image x of the previous round of iteration. i Perform preliminary denoising to obtain the intermediate generated image of this round of iteration The specific processing process is expressed as:

[0111]

[0112] In formula (2), σ i is the noise term of the previous iteration; σ i-1 is the noise term of this iteration; z means it satisfies the standard normal distribution Noise; I is a constant, for example, the value is 1; is an intermediate variable. Formula (2) is used to characterize the preliminary denoising process using a numerical stochastic differential equation solver.

[0113] In this way, the image generated by the previous round of iteration can be initially denoised for subsequent processing.

[0114] In some possible implementations, in this round of iteration (i-1th round), the noise term σ corresponding to this round of iteration may be further calculated. i-1 , generating a normal distribution Noise Noise the candidate trend graph x to generate the noisy image y of this iteration, satisfying The noisy image y is used for subsequent correction operations.

[0115] In some possible implementations, in the correction operation of this round of iteration (i-1th round), according to the noisy image and mask image area m of this round of iteration, the intermediate generated image of this round of iteration is denoised, reconstructed and consistency corrected through the image generation model to obtain the generated image of this round of iteration.

[0116] In some possible implementations, the correction operation includes multiple processing, and the number of processing times T is the number of annealing Langevin kinetic sampling times. Each processing includes:

[0117] The rectified image processed last time is subjected to denoising and reconstruction processing to obtain the reconstructed image processed this time; wherein the rectified image processed last time corresponding to the first processing is the intermediate generated image of this round of iteration;

[0118] According to the noise-added image of this round of iteration and the mask image area, consistency correction is performed on the reconstructed image processed this time to obtain a corrected image processed this time;

[0119] When this round of processing is the last processing, the corrected image processed this time is used as the generated image of the current round of iteration.

[0120] For example, for the t-th processing (1≤t≤T), the annealing Langevin dynamics processing method is used to generate the image of the previous processing correction image. Perform denoising and reconstruction processing to obtain the reconstructed image processed this time It is expressed as:

[0121]

[0122] In formula (3), α i-1 Denotes the noise term σ i-1 The corresponding step size; α i Denotes the noise term σ i The corresponding step size; zt It means that it satisfies the standard normal distribution The first processing corresponds to the corrected image of the previous processing. Generate the intermediate image for this round of iteration That is, the processing result of the previous estimation operation.

[0123] In some possible implementations, the reconstructed image processed this time can be processed according to the noise-added image y and the mask image region m of this iteration. Perform consistency correction to obtain the corrected image processed this time, which is expressed as:

[0124]

[0125] Among them, formula (4) is used to represent the consistency correction of the reconstructed image, update the reconstructed image, so that the non-mask image area (that is, the image area corresponding to the historical time period) remains unchanged, and only the part corresponding to the mask image area m in the reconstructed image is retained, so as to avoid the generated image from deviating from the historical business trend, thereby improving the accuracy of the generated image. After correction and update, the corrected image of this processing is obtained.

[0126] In this way, the above processing method is used for T times of processing. When this processing is the Tth processing (the last processing), the corrected image processed this time is As the final generated image of this round of iteration That is, adopt renew

[0127] In this way, the loss of original data can be effectively avoided and the accuracy of image restoration can be improved.

[0128] In some possible implementations, the above processing method is used to perform T times of one estimation operation and one correction operation in each round of iteration, which can continuously improve the accuracy of the generated image.

[0129] When this round of iteration is the Lth round of iteration (the last round of iteration), the generated image of this round of iteration can be As the final predicted business trend graph, the whole denoising generation process is completed. The predicted business trend graph includes a first image area corresponding to the historical time period and a second image area corresponding to the predicted time period.

[0130] Figure 3b A schematic diagram of a predicted business trend diagram of a business prediction method provided in an embodiment of the present disclosure. Figure 3a Similar to, Figure 3bThe horizontal axis represents the number of statistical units, such as weeks; the vertical axis represents the numerical range corresponding to the standardized indicator data. The historical time period corresponds to weeks 0 to 92, and the forecast time period corresponds to the 12 weeks after the historical time period, that is, weeks 93 to 104.

[0131] like Figure 3b As shown in the forecast business trend graph, the left side is the first image area corresponding to the historical time period (0 to 92 weeks), and the right side is the second image area corresponding to the forecast time period (93 to 104 weeks). In the example, only the business indicators with a value interval of [0.4, 0.6] are denoised to obtain the curve of the business indicator in the forecast time period.

[0132] In some possible implementations, after obtaining the predicted business trend graph, the business trend of the target business may be predicted according to the predicted business trend graph in step S23. Step S23 may include:

[0133] The demand description information of the target business and the predicted business trend graph are subjected to prediction processing to obtain a business prediction result of the target business.

[0134] For example, a multimodal big model can be combined for business prediction. The multimodal big model can be any language-based big model based on GPT, Bert, etc. The present disclosure does not limit the specific type of the multimodal big model.

[0135] In some possible implementations, based on specific business needs (such as application trends for business resources, trends in user consumption needs, etc.), demand description information to be input into the multimodal large model can be prepared in advance, for example, "Please describe the final trend of the broken line with a numerical range of [0.4, 0.6] in the figure, and give a conclusion: rising or falling."

[0136] In some possible implementations, the demand description information and the predicted business trend chart can be simultaneously input into a multimodal large model for processing, the business demand can be analyzed and a final conclusion can be given, and the final business forecast result can be output, for example, the application trend of business resources is rising.

[0137] In some possible implementations, the business forecast results can also be determined in other ways. For example, the demand description information can be combined, such as "determine the final trend direction of the broken line with a value interval of [0.4, 0.6] in the graph", and the curves of the historical time period and the forecast time period in the forecast business trend graph can be compared. Figure 3bCompare the curve A1 of the adjacent part of the historical time period and the predicted time period (weeks 81-92) with the curve A2 of the predicted time period (weeks 93-104) to check the changing trend of the curve. If the curve A2 is lower than the curve A1, it can be determined that the business forecast result is decreasing; conversely, if the curve A2 is higher than the curve A1, it can be determined that the business forecast result is increasing. The present disclosure does not limit the specific method of determining the business forecast result.

[0138] In this way, the advantages of multimodal large models can be combined to obtain business forecast results from the reconstructed forecast business trend chart, so as to give clearer business forecast results based on different business needs, thereby improving the readability and clarity of business forecast results and further improving the effect of business forecasting.

[0139] For time series prediction tasks, related technologies tend to start from data at a time point and use sequence data as model input for prediction. The data at each time point will have a greater impact on the prediction results. Therefore, the sequence model is more sensitive to missing data samples. In addition, the prediction time step and input time series length will affect the final performance of the model.

[0140] The disclosed embodiments can effectively alleviate the above problems. Compared with the time series prediction method in the related art that uses sequence data as model input, the business prediction method according to the disclosed embodiment plots the existing user historical data into a historical business trend graph, generates a predicted business trend graph through an image generation model, and predicts and analyzes business needs in an image processing manner, that is, realizes the prediction trend line extension task based on image restoration, thereby improving the effect of business prediction.

[0141] Moreover, in the historical data, if the data of a certain time point (segment) is missing, the missing data is filled by curve fitting and image fitting to ensure the consistency of sample data distribution. Subsequently, the drawn image is used as a sample for model training, and the distribution of the overall data in the image is learned by the diffusion model, so as to fully capture the specificity and relevance of different historical behaviors of users, and is not easily affected by the length of the time series.

[0142] According to the business prediction method of the embodiment of the present disclosure, it is possible to achieve prediction of data of different time steps by adjusting the size of the prediction time period, thereby overcoming the problem of inaccurate prediction of long step lengths in the time series prediction model in the related art. In addition, the embodiment of the present disclosure designs a mask that can be directly associated with the prediction result, and uses the diffusion model to complete the denoising generation of the image, thereby improving the efficiency, accuracy and flexibility of the prediction.

[0143] According to the business forecasting method of the embodiment of the present disclosure, it is also possible to combine the advantages of the multimodal big model and use the multimodal big model to predict specific tasks on the predicted business trend chart and business needs, so as to output business forecasting results for the target business, thereby giving clearer business forecasting results based on different business needs, improving the readability and clarity of the business forecasting results, and further improving the effect of business forecasting.

[0144] It can be understood that the above-mentioned various method embodiments mentioned in the present disclosure can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, the present disclosure will not repeat them. It can be understood by those skilled in the art that in the above-mentioned method of the specific implementation method, the specific execution order of each step should be determined according to its function and possible internal logic.

[0145] In addition, the present disclosure also provides a business prediction device, an electronic device, and a computer-readable storage medium, all of which can be used to implement any business prediction method provided by the present disclosure. The corresponding technical solutions and descriptions are referred to in the corresponding records of the method part and will not be repeated here.

[0146] Figure 5 A block diagram of a business prediction device provided in an embodiment of the present disclosure.

[0147] Reference Figure 5 , the embodiment of the present disclosure provides a business prediction device, the business prediction device comprising:

[0148] A historical trend determination module 51 is used to determine a historical business trend graph of the target business according to the historical indicator data of the business indicator of the target business within a historical time period;

[0149] A forecast trend generating module 52, configured to generate a forecast business trend graph of the target business according to the historical business trend graph and the forecast time period;

[0150] The result determination module 53 is used to predict the business trend of the target business according to the predicted business trend graph to obtain the business prediction result of the target business.

[0151] In some possible implementations, the forecast trend generation module 52 is used to: generate a candidate trend graph based on the historical business trend graph and the forecast time period; and denoise the candidate trend graph based on the noise sequence to obtain a forecast business trend graph of the target business.

[0152] In some possible implementations, the noise sequence includes L noise items, and the denoising generation process includes L rounds of iterations, one round of iteration corresponds to one noise item; the candidate trend graph includes a mask image area corresponding to the prediction time period; the prediction trend generation module 52 is used to: denoise the generated image of the previous round of iteration to obtain an intermediate generated image of the current round of iteration; wherein the generated image of the previous round of iteration of the first round of iteration is the noisy image of the candidate trend graph; determine the noisy image of the current round of iteration based on the noise item corresponding to the current round of iteration and the candidate trend graph; denoise and reconstruct and perform consistency correction on the intermediate generated image of the current round of iteration based on the noisy image of the current round of iteration and the mask image area to obtain the generated image of the current round of iteration; when the current round of iteration is the Lth round of iteration, the generated image of the current round of iteration is used as the predicted business trend graph.

[0153] In some possible implementations, the prediction trend generation module 52 is used to: adopt a preset calculation method to denoise the generated image of the previous round of iteration through the noise terms of the previous round of iteration and the noise terms of the current round of iteration in the noise sequence to obtain the intermediate generated image of the current round of iteration.

[0154] In some possible implementations, denoising, reconstructing and consistency correction of the intermediate generated image of this round of iteration includes multiple processing, and each processing includes: denoising and reconstructing the corrected image of the previous processing to obtain the reconstructed image of this processing; wherein the corrected image of the previous processing corresponding to the first processing is the intermediate generated image of this round of iteration; according to the noisy image of this round of iteration and the mask image area, consistency correction is performed on the reconstructed image of this processing to obtain the corrected image of this processing; when this round of processing is the last processing, the corrected image of this processing is used as the generated image of this round of iteration.

[0155] In some possible implementations, the historical trend determination module 51 is used to: standardize the historical indicator data of each business indicator to obtain standardized indicator data of the business indicator; wherein the standardized indicator data of different business indicators are in different numerical ranges; draw a historical business trend chart based on the standardized indicator data of the business indicator; wherein the historical business trend chart includes a curve of the indicator data changes of each business indicator within the historical time period.

[0156] In some possible implementations, the result determination module 53 is used to: perform forecasting processing on the demand description information of the target business and the forecasted business trend graph to obtain a business forecast result of the target business.

[0157] Each module in the above-mentioned business prediction device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module above.

[0158] Figure 6 A block diagram of an electronic device provided in an embodiment of the present disclosure.

[0159] Reference Figure 6 An embodiment of the present disclosure provides an electronic device, which includes: at least one processor 701; at least one memory 702, and one or more I / O interfaces 703, connected between the processor 701 and the memory 702; wherein the memory 702 stores one or more computer programs that can be executed by the at least one processor 701, and the one or more computer programs are executed by the at least one processor 701 so that the at least one processor 701 can execute the above-mentioned business prediction method.

[0160] Each module in the above electronic device can be implemented in whole or in part by software, hardware, or a combination thereof. Each module can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to each module.

[0161] The disclosed embodiments also provide a computer-readable storage medium, which may be a volatile or non-volatile computer-readable storage medium, on which a computer program is stored, wherein when the computer program is executed by a processor: based on historical indicator data of business indicators of the target business within a historical time period, a historical business trend graph of the target business is determined; based on the historical business trend graph and a predicted time period, a predicted business trend graph of the target business is generated; based on the predicted business trend graph, a business trend of the target business is predicted to obtain a business prediction result of the target business.

[0162] In some possible implementations, the processor generates a predicted service trend graph of the target service according to the historical service trend graph and the predicted time period, including:

[0163] Generate a candidate trend graph based on the historical business trend graph and the forecast time period;

[0164] According to the noise sequence, the candidate trend graph is subjected to denoising and generation processing to obtain a predicted business trend graph of the target business.

[0165] In some possible implementations, the noise sequence includes L noise items, the denoising generation process includes L rounds of iterations, and one round of iteration corresponds to one noise item; the candidate trend graph includes a mask image area corresponding to the prediction time period;

[0166] The processor performs denoising and generating processing on the candidate trend graph according to the noise sequence to obtain the predicted business trend graph of the target business, including: denoising and generating the generated image of the previous round of iteration to obtain the intermediate generated image of the current round of iteration; wherein the generated image of the previous round of iteration of the first round of iteration is the noise-added image of the candidate trend graph;

[0167] According to the noise item corresponding to this round of iteration and the candidate trend chart, the noisy image of this round of iteration is determined; according to the noisy image of this round of iteration and the mask image area, the intermediate generated image of this round of iteration is denoised, reconstructed and consistency corrected to obtain the generated image of this round of iteration; when this round of iteration is the Lth round of iteration, the generated image of this round of iteration is used as the predicted business trend chart.

[0168] In some possible implementations, the processor denoises the generated image of the previous round of iterations to obtain an intermediate generated image of the current round of iterations, including: using a preset calculation method to denoise the generated image of the previous round of iterations through the noise terms of the previous round of iterations and the noise terms of the current round of iterations in the noise sequence to obtain the intermediate generated image of the current round of iterations.

[0169] In some possible implementations, the processor performs denoising, reconstruction and consistency correction on the intermediate generated image of this round of iteration, including multiple processing, and each processing includes: performing denoising and reconstruction processing on the corrected image of the previous processing to obtain the reconstructed image of this processing; wherein the corrected image of the previous processing corresponding to the first processing is the intermediate generated image of this round of iteration; according to the noisy image of this round of iteration and the mask image area, performing consistency correction on the reconstructed image of this processing to obtain the corrected image of this processing; when this round of processing is the last processing, the corrected image of this processing is used as the generated image of this round of iteration.

[0170] In some possible implementations, the processor determines a historical business trend graph of the target business based on historical indicator data of the business indicators of the target business within a historical time period, including: standardizing the historical indicator data of each business indicator to obtain standardized indicator data of the business indicator; wherein the standardized indicator data of different business indicators are in different numerical ranges; drawing a historical business trend graph based on the standardized indicator data of the business indicators; wherein the historical business trend graph includes a curve of indicator data changes of each business indicator within the historical time period.

[0171] In some possible implementations, the processor predicts the business trend of the target business based on the predicted business trend graph to obtain a business prediction result of the target business, including: performing prediction processing on the demand description information of the target business and the predicted business trend graph to obtain a business prediction result of the target business.

[0172] The embodiments of the present disclosure also provide a computer program product, including a computer-readable code, or a non-volatile computer-readable storage medium carrying the computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the above-mentioned business prediction method.

[0173] It will be appreciated by those skilled in the art that all or some of the steps, systems, and functional modules / units in the methods disclosed above may be implemented as software, firmware, hardware, and appropriate combinations thereof. In a hardware implementation, the division between the functional modules / units mentioned in the above description does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed by several physical components in cooperation. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor, or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application-specific integrated circuit. Such software may be distributed on a computer-readable storage medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium).

[0174] As is known to those of ordinary skill in the art, the term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable program instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technology, portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and can be accessed by a computer. In addition, it is known to those of ordinary skill in the art that communication media typically contain computer-readable program instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transmission mechanism, and may include any information delivery medium.

[0175] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0176] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0177] The computer program product described herein may be implemented in hardware, software, or a combination thereof. In one optional embodiment, the computer program product is embodied as a computer storage medium, and in another optional embodiment, the computer program product is embodied as a software product, such as a software development kit (SDK), etc.

[0178] Various aspects of the present disclosure are described herein with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer-readable program instructions.

[0179] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, thereby producing a machine, so that when these instructions are executed by the processor of the computer or other programmable data processing device, a device that implements the functions / actions specified in one or more boxes in the flowchart and / or block diagram is generated. These computer-readable program instructions can also be stored in a computer-readable storage medium, and these instructions cause the computer, programmable data processing device, and / or other equipment to work in a specific manner, so that the computer-readable medium storing the instructions includes a manufactured product, which includes instructions for implementing various aspects of the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0180] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operating steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to implement the functions / actions specified in one or more boxes in the flowchart and / or block diagram.

[0181] The flow chart and block diagram in the accompanying drawings show the possible architecture, function and operation of the system, method and computer program product according to multiple embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a part of a module, program segment or instruction, and a part of the module, program segment or instruction includes one or more executable instructions for realizing the specified logical function. In some alternative implementations, the function marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two continuous square boxes can actually be executed substantially in parallel, and they can sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or action, or can be implemented with a combination of special hardware and computer instructions.

[0182] Example embodiments have been disclosed herein, and although specific terms are employed, they are used and should be interpreted only in a general illustrative sense and not for limiting purposes. In some instances, it will be apparent to those skilled in the art that, unless otherwise expressly noted, features, characteristics, and / or elements described in conjunction with a particular embodiment may be used alone or in combination with features, characteristics, and / or elements described in conjunction with other embodiments. Therefore, those skilled in the art will appreciate that various changes in form and detail may be made without departing from the scope of the present disclosure as set forth in the appended claims.

Claims

1. A business forecasting method, It is characterized in that include: Determine a historical business trend graph of the target business based on historical indicator data of the target business within a historical time period; Generate a forecasted business trend graph of the target business according to the historical business trend graph and the forecast time period; According to the predicted business trend graph, the business trend of the target business is predicted to obtain a business prediction result of the target business.

2. The method according to claim 1, It is characterized in that The generating the predicted business trend graph of the target business according to the historical business trend graph and the predicted time period includes: Generate a candidate trend graph based on the historical business trend graph and the forecast time period; According to the noise sequence, the candidate trend graph is subjected to denoising and generation processing to obtain a predicted business trend graph of the target business.

3. The method according to claim 2, It is characterized in that The noise sequence includes L noise items, the denoising generation process includes L rounds of iterations, and one round of iteration corresponds to one noise item; the candidate trend graph includes a mask image area corresponding to the prediction time period; The step of performing denoising and generating processing on the candidate trend graph according to the noise sequence to obtain a predicted business trend graph of the target business includes: De-noising the generated image of the previous round of iteration to obtain an intermediate generated image of the current round of iteration; wherein the generated image of the previous round of iteration of the first round of iteration is the noisy image of the candidate trend graph; Determining a noisy image for this round of iteration according to the noise item corresponding to this round of iteration and the candidate trend graph; According to the noisy image of this round of iteration and the mask image area, denoising, reconstructing and consistency correction are performed on the intermediate generated image of this round of iteration to obtain the generated image of this round of iteration; When the current iteration is the Lth iteration, the generated image of the current iteration is used as the predicted business trend graph.

4. The method according to claim 3, It is characterized in that The step of denoising the generated image of the previous round of iteration to obtain the intermediate generated image of the current round of iteration includes: By adopting a preset operation method, the generated image of the previous round of iteration is denoised through the noise term of the previous round of iteration and the noise term of the current round of iteration in the noise sequence, so as to obtain the intermediate generated image of the current round of iteration.

5. The method according to claim 3, It is characterized in that The denoising, reconstruction and consistency correction of the intermediate generated image of this round of iteration includes multiple processing, each of which includes: The rectified image processed last time is subjected to denoising and reconstruction processing to obtain the reconstructed image processed this time; wherein the rectified image processed last time corresponding to the first processing is the intermediate generated image of this round of iteration; According to the noise-added image of this round of iteration and the mask image area, consistency correction is performed on the reconstructed image processed this time to obtain a corrected image processed this time; When this round of processing is the last processing, the corrected image processed this time is used as the generated image of the current round of iteration.

6. The method according to claim 1, It is characterized in that Determining the historical business trend graph of the target business according to the historical indicator data of the business indicator of the target business within the historical time period includes: Standardizing the historical indicator data of each business indicator to obtain standardized indicator data of the business indicator; wherein the standardized indicator data of different business indicators are in different value ranges; A historical business trend graph is drawn based on the standardized indicator data of the business indicator; wherein the historical business trend graph includes a curve of indicator data changes of each business indicator within the historical time period.

7. The method according to claim 1, It is characterized in that The predicting the business trend of the target business according to the predicted business trend graph to obtain the business prediction result of the target business includes: The demand description information of the target business and the predicted business trend graph are subjected to prediction processing to obtain a business prediction result of the target business.

8. A business prediction device, It is characterized in that include: A historical trend determination module, used to determine a historical business trend graph of the target business based on the historical indicator data of the business indicator of the target business within a historical time period; A forecast trend generation module, used to generate a forecast business trend graph of the target business based on the historical business trend graph and the forecast time period; The result determination module is used to predict the business trend of the target business according to the predicted business trend graph to obtain the business prediction result of the target business.

9. An electronic device, It is characterized in that include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores one or more computer programs that can be executed by the at least one processor, and the one or more computer programs are executed by the at least one processor to enable the at least one processor to execute the business forecasting method according to any one of claims 1 to 7.

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 computer program implements the business prediction method according to any one of claims 1 to 7.

11. A computer program product, It is characterized in that It includes a computer-readable code, or a non-volatile computer-readable storage medium carrying a computer-readable code. When the computer-readable code runs in a processor of an electronic device, the processor in the electronic device executes the business prediction method as described in any one of claims 1 to 7.

Images