A data processing method, device, apparatus, and storage medium

Abstract

This application discloses a data processing method, apparatus, device, and storage medium. The solution includes: determining an input reference value based on a first input value; the input reference value characterizing the magnitude of a first output value corresponding to the first input value; determining a deviation based on a second output value and a second actual value; determining a first state value based on the input reference value, the deviation, and a second state value; the second state value being the state value corresponding to the second input value; and determining a first output value based at least on the first state value. This approach, on the one hand, adds a verification process for calculating the deviation during data analysis, achieving automatic verification and simplifying implementation; on the other hand, in determining the output value, it considers both the influence of the input value and the influence of historical state values, thus improving the accuracy of the prediction results.

Description

Technical Field

[0001] This application relates to the field of data processing technology, including but not limited to data processing methods, apparatus, devices, and storage media. Background Technology

[0002] With the rapid development of computer technology, more and more technologies are being applied in the financial sector, and the traditional financial industry is gradually transforming into Fintech. However, due to the security and real-time requirements of the financial industry, higher demands are being placed on technology. Among these, data analysis is playing an increasingly important role in business practice. Currently, data analysis processes involve large sample sizes, relatively complex calculations, and the data sampling dimensions are mostly determined by experienced business personnel. This results in excessive human intervention in the data analysis results, and a lack of stable correlation between input and output items.

[0003] For example, when performing data analysis, related technologies use data models to analyze input values ​​and obtain prediction results; then determine the final result based on the prediction results manually; and finally verify the prediction results manually.

[0004] Thus, on the one hand, since the prediction results and the verification process are separate, the accuracy of the prediction results obtained from the data model is poor. On the other hand, the verification is carried out manually, which makes the process complex, time-consuming and labor-intensive. Summary of the Invention

[0005] This application provides a data processing method, apparatus, device, and storage medium. Regarding the solution of this application, on the one hand, a deviation calculation verification process is added during the data analysis process to achieve automatic verification, which is simple to implement; on the other hand, in the process of determining the output value, both the influence of the input value and the influence of historical state values ​​are considered, thus improving the accuracy of the prediction results.

[0006] The technical solution of this application is implemented as follows:

[0007] This application provides a data processing method, the method comprising:

[0008] Based on the first input value, an input reference value is determined; the input reference value is used to characterize the magnitude of the first output value that affects the first input value.

[0009] Based on the second output value and the second actual value, the deviation is determined; the deviation is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value;

[0010] Based on the input reference value, the deviation, and the second state value, a first state value is determined; the second state value is the state value corresponding to the second input value.

[0011] The first output value is determined based at least on the first state value.

[0012] This application provides a data processing apparatus, the apparatus comprising:

[0013] The first determining unit is configured to determine an input reference value based on a first input value; the input reference value is used to characterize the magnitude of the first output value that affects the first input value.

[0014] The second determining unit is used to determine the deviation degree based on the second output value and the second actual value; the deviation degree is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value;

[0015] The third determining unit is used to determine a first state value based on the input reference value, the deviation, and the second state value; the second state value is the state value corresponding to the second input value.

[0016] The fourth determining unit is used to determine the first output value based at least on the first state value.

[0017] This application also provides an electronic device, including: a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the program to implement the above-described data processing method.

[0018] This application also provides a storage medium on which a computer program is stored, which, when executed by a processor, implements the above-described data processing method.

[0019] The data processing method, apparatus, device, and storage medium provided in this application include: determining an input reference value based on a first input value; the input reference value being used to characterize the magnitude of a first output value corresponding to the first input value; determining a deviation degree based on a second output value and a second actual value; the deviation degree being used to characterize the degree of deviation between the second output value and the second actual value; the second output value being the output value corresponding to the second input value, the second actual value being the actual value corresponding to the second input value, and the second input value being a historical input value; determining a first state value based on the input reference value, the deviation degree, and a second state value; the second state value being the state value corresponding to the second input value; and determining a first output value based at least on the first state value. Regarding the solution of this application, firstly, during data analysis, the influence of the deviation degree on the output result is considered in the process of obtaining the first output value, which is equivalent to adding a verification process for calculating the deviation degree during the data analysis process, realizing an automatic verification function, and simplifying implementation; secondly, in the process of determining the output value, both the influence of the input value and the influence of historical state values ​​are considered, thus improving the accuracy of the prediction result. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of an optional structure of the data processing system provided in an embodiment of this application;

[0021] Figure 2 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0022] Figure 3 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0023] Figure 4 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0024] Figure 5 A schematic flowchart of an optional data processing method provided in an embodiment of this application;

[0025] Figure 6 A schematic diagram of an optional data processing procedure provided for an embodiment of this application;

[0026] Figure 7 This is a schematic diagram of an optional structure of a recurrent neural network provided in an embodiment of this application;

[0027] Figure 8 This is a schematic diagram of an optional structure of the data processing apparatus provided in an embodiment of this application;

[0028] Figure 9 This is a schematic diagram of an optional structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the specific technical solutions of the application will be further described in detail below with reference to the accompanying drawings of the embodiments of this application. The following embodiments are used to illustrate this application, but are not intended to limit the scope of this application.

[0030] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0031] In the following description, the terms "first," "second," and "third" are used only to distinguish different objects and do not represent a specific order of objects, nor are they constituting a chronological order. It is understood that "first," "second," and "third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0033] This application provides data processing methods, apparatus, devices, and storage media. In practical applications, the data processing method can be implemented by a data processing apparatus, and the functional entities in the data processing apparatus can be collaboratively implemented by the hardware resources of electronic devices, such as computing resources like processors and communication resources (such as those used to support various communication methods like optical fiber and cellular networks).

[0034] The data processing method provided in this application embodiment is applied to a data processing system, which includes a data processing terminal.

[0035] Specifically, the data processing end performs the following steps: Based on a first input value, determine an input reference value; the input reference value is used to characterize the magnitude of the first output value corresponding to the first input value; Based on a second output value and a second actual value, determine a deviation degree; the deviation degree is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value; Based on the input reference value, the deviation degree, and a second state value, determine a first state value; the second state value is the state value corresponding to the second input value; Based at least on the first state value, determine a first output value.

[0036] Optionally, the data processing system may also include a data acquisition terminal, etc.

[0037] The data acquisition terminal is used to collect historical input values, historical output values, etc.

[0038] As an example, the structure of the data processing system 10 can be as follows: Figure 1 As shown, it includes: a data processing terminal 101 and a data acquisition terminal 102; the data processing terminal 101 and the data acquisition terminal 102 can communicate.

[0039] The data processing terminal 101 is used to perform the following: determining an input reference value based on a first input value; the input reference value is used to characterize the magnitude of the first output value corresponding to the first input value; determining a deviation based on a second output value and a second actual value; the deviation is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value; determining a first state value based on the input reference value, the deviation, and a second state value; the second state value is the state value corresponding to the second input value; and determining a first output value based at least on the first state value.

[0040] The data processing terminal 101 can be an electronic device with relevant data processing capabilities, such as a server or computer. For example, the data processing terminal 101 can be a physical service, a physical computer, a cloud computer, or a cloud server.

[0041] The data acquisition terminal 102 is used to display various information of the first device. For example, the data acquisition terminal 102 can display information of the first device in a first state or information of the first device in a second state.

[0042] The data acquisition terminal 102 is used to collect historical input values, historical output values, etc., and can be an electronic device with relevant data processing capabilities.

[0043] The data processing terminal 101 and the data acquisition terminal 102 can communicate via wired or wireless communication methods. This application embodiment does not specifically limit this and can be configured according to actual needs.

[0044] It should be noted that the data processing terminal 101 and the data acquisition terminal 102 mentioned above can be deployed in the same device in an integrated manner, or they can be deployed independently in different devices. This application embodiment does not limit this.

[0045] Below, in conjunction with Figure 1 The schematic diagram of the data processing system shown illustrates various embodiments of the data processing methods, apparatus, devices, and storage media provided in this application.

[0046] In a first aspect, embodiments of this application provide a data processing method, which is applied to a data processing apparatus; wherein the data processing apparatus can be deployed in... Figure 1 The data processing terminal 101 is described below. The data processing procedure provided in the embodiments of this application will be explained below.

[0047] Figure 2 This diagram illustrates a possible data processing method, such as... Figure 2 As shown, this data processing method may include, but is not limited to, [other methods]. Figure 2 S201 to S204 are shown.

[0048] S201. The data processing end determines the input reference value based on the first input value.

[0049] Input values ​​can be referred to as independent variables. This application does not limit the specific content of the input values, which can be configured according to actual needs. For example, input values ​​may include one or more of the following: daily channel advertising volume, daily user pageviews, daily user insurance clicks, and daily user payment clicks.

[0050] The output value can be referred to as the dependent variable. This application does not limit the specific content of the output value; it can be configured according to actual needs. For example, the output value can be daily sales revenue.

[0051] The first input value is any one of the input values ​​in this data processing. In this embodiment, the number of input values ​​processed by the data processing terminal at one time is not specifically limited and can be configured according to actual needs. For example, the data processing terminal can process only one input value at a time; correspondingly, the first input value is that input value. As another example, the data processing terminal can process multiple input values ​​at one time; correspondingly, the first input value is any one of those multiple input values.

[0052] Input a reference value, which is used to characterize the magnitude of the first output value that affects the first input value.

[0053] In this embodiment of the application, the specific method by which the data processing end determines the input reference value based on the first input value is not limited, and can be configured according to actual needs.

[0054] In one possible implementation, the data processing end can determine the first input value as the input reference value.

[0055] In another possible implementation, the data processing end can perform a calculation on the first input value to obtain the magnitude of the first output value that affects the first input value, and determine the magnitude as the input reference value.

[0056] S202. The data processing end determines the deviation based on the second output value and the second actual value.

[0057] Wherein, the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value. This application embodiment does not limit the specific value of the second input value and can configure it according to actual needs.

[0058] In one possible implementation, the second input value can be any one of the previous input values, and correspondingly, the second output value is the output value corresponding to any one of the previous input values, and the second actual value is the actual value corresponding to any one of the previous input values; or, the second input value can also be the average input value among the previous input values, and correspondingly, the second output value is the output value corresponding to the average input value, and the second actual value is the actual value corresponding to the average input value.

[0059] In another possible implementation, the second input value can also be one of the previous N input values, where N is greater than 1. Correspondingly, the second output value is the output value corresponding to the selected input value among the previous N input values, and the second actual value is the actual value corresponding to the selected input value among the previous N input values.

[0060] Deviation degree is used to characterize the degree of deviation between the second output value and the second actual value.

[0061] The embodiments of this application do not limit the specific method by which the data processing end determines the deviation based on the second output value and the second actual value, and can be configured according to actual needs.

[0062] In one possible implementation, the data processing end can determine the deviation as the difference between the second output value and the second actual value.

[0063] In one possible implementation, the data processing end can determine the deviation as the variance between the second output value and the second actual value.

[0064] Understandably, there are many other ways to determine deviation, such as standard deviation, which will not be listed here.

[0065] S203. The data processing terminal determines the first state value based on the input reference value, the deviation, and the second state value.

[0066] The second state value is the state value corresponding to the second input value.

[0067] This application embodiment does not limit the specific method by which the data processing end determines the first state value based on the input reference value, deviation, and second state value, and can be configured according to actual needs. For example, the data processing end can perform a calculation on the input reference value, deviation, and second state value to obtain the first state value.

[0068] Optionally, the data processing end can also save the first state value.

[0069] S204. The data processing end determines the first output value based at least on the first state value.

[0070] The embodiments of this application do not impose a unique limitation on the specific method or reference parameters for the data processing end to determine the first output value, and can be configured according to actual needs.

[0071] In one possible implementation, the data processing end can determine the first output value based on the first state value.

[0072] In another possible implementation, the data processing unit can determine the first output value based on the first state value and other parameters. For example, the data processing unit can determine the first output value based on the first state value and the correlation coefficient.

[0073] It should be noted that the data processing end can determine the first output value using a recurrent neural network model. In other words, S201 to S204 described above can be executed using a recurrent neural network model. It is understood that the data processing end can also determine the first output value using other data analysis models, and this application embodiment does not limit this to a single method.

[0074] The data processing method provided in this application includes: determining an input reference value based on a first input value; the input reference value being used to characterize the magnitude of the first output value corresponding to the first input value; determining a deviation degree based on a second output value and a second actual value; the deviation degree being used to characterize the degree of deviation between the second output value and the second actual value; the second output value being the output value corresponding to the second input value, the second actual value being the actual value corresponding to the second input value, and the second input value being a historical input value; determining a first state value based on the input reference value, the deviation degree, and a second state value; the second state value being the state value corresponding to the second input value; and determining a first output value based at least on the first state value. Regarding the solution of this application, firstly, during data analysis, the influence of the deviation degree on the output result is considered in the process of obtaining the first output value, which is equivalent to adding a verification process for calculating the deviation degree during the data analysis process, realizing an automatic verification function, and simplifying implementation; secondly, in the process of determining the output value, both the influence of the input value and the influence of historical state values ​​are considered, thus improving the accuracy of the prediction result.

[0075] The following describes the process by which the S201 data processing terminal determines the input reference value based on the first input value.

[0076] In one possible implementation, the data processing end can determine the input reference value according to the first formula.

[0077] The first formula includes: The F(x) nt ) represents the input reference value; the x nt This represents the first input value; the This represents the average value of the input.

[0078] In this way, by determining the input reference value through this possible implementation method, the accuracy of the obtained input reference value is relatively high, which can further improve the accuracy of the prediction results.

[0079] Understandably, the data processing end can also determine the input reference value based on other formulas. This application embodiment does not limit this and can be configured according to actual needs. For example, a new formula can be obtained by equivalently replacing some content in the first formula. Exemplarily, it can also be based on... To determine the input reference value.

[0080] The following explains the process by which the S202 data processing terminal determines the deviation based on the second output value and the second actual value.

[0081] In one possible implementation, the data processing end can determine the deviation according to the second formula.

[0082] The second formula includes: The C nt Indicates the deviation; the y nt This represents the second actual value; the f nt ' indicates the second output value.

[0083] In this way, by determining the deviation through this possible implementation method, the accuracy of the obtained deviation is relatively high, which can further improve the accuracy of the prediction results.

[0084] Understandably, the data processing end can also determine the deviation based on other formulas. This application embodiment does not limit this, and configuration can be made according to actual needs. For example, a new formula can be obtained by equivalent substitution of some content in the second formula. Exemplarily, it can also be based on... To determine the input reference value.

[0085] The process by which the S203 data processing terminal determines the first state value based on the input reference value, the deviation, and the second state value will be described below.

[0086] In one possible implementation, the data processing end can determine the first state value according to the third formula;

[0087] The third formula includes: S nt =[(1-C nt )×S n(t-1) +F(x nt )]; The S nt This represents the first state value; the C nt The deviation is represented by F(x). nt ) represents the input reference value, and the S n(t-1) This represents the second state value.

[0088] In this way, by determining the first state value through this possible implementation method, the accuracy of the first state value obtained is relatively high, which can further improve the accuracy of the prediction results.

[0089] Understandably, the data processing end can also determine the first state value based on other formulas. This application embodiment does not limit this, and configuration can be made according to actual needs. For example, a new formula can be obtained by equivalent substitution of some content in the third formula. Exemplarily, it can also be determined based on S... nt =[(1-C nt )×S n(t-1) +MF(x ntThe input reference value is determined by [M]. The value of M can be configured according to actual needs.

[0090] The process by which the S204 data processing terminal determines the first output value based at least on the first state value will be described below. This process may include, but is not limited to, method 1 or method 2 described below.

[0091] Method 1: Determine the first output value based on the first state value;

[0092] Method 2: Determine the first output value based on the first state value and the correlation coefficient.

[0093] For method 1, the data processing end can directly determine the first state value as the first output value.

[0094] In this way, the implementation of this method is simple and easy to implement.

[0095] The following explains the process of determining the first output value based on the first state value and the correlation coefficient in Method 2. Figure 3 As shown, the process may include, but is not limited to, S301 to S303 described below.

[0096] S301, The data processing end obtains the historical output set and multiple types of historical input values.

[0097] The historical output set includes multiple historical output values, which are the actual historical output values ​​collected. This application embodiment does not limit the number of historical output values ​​included in the historical output set, nor the method of acquisition; it can be configured according to actual needs.

[0098] In one possible implementation, the data processing terminal itself collects multiple actual historical output values ​​as a historical output set.

[0099] In another possible implementation, the data processing end can receive a set of historical outputs sent by the data acquisition end.

[0100] This application does not limit the specific types of historical input values ​​involved in the various types of historical input values, and can be configured according to actual needs. For example, the specific types of historical input values ​​involved in the various types of historical input values ​​may include: daily channel advertising volume, daily user page views, daily user insurance clicks, and daily user payment clicks.

[0101] This application does not limit the methods for obtaining various types of historical input values, and can be configured according to actual needs.

[0102] In one possible implementation, the data processing terminal itself collects multiple types of historical input values.

[0103] In another possible implementation, the data processing end can receive multiple types of historical input values ​​sent by the data acquisition end.

[0104] S302. The data processing end calculates the correlation coefficient between the input and the output based on the historical output set and the historical input values ​​of the multiple types.

[0105] The correlation coefficient is used to characterize the degree of association between input and output values. This application embodiment does not limit the specific method by which the data processing end calculates the correlation coefficient between input and output; it can be configured according to actual needs. For example, the data processing end can perform correlation calculations on historical output sets and multiple types of historical input values ​​to obtain the correlation coefficient between input and output.

[0106] S303. The data processing end determines the first output value based on the first state value and the correlation coefficient.

[0107] In one possible implementation, the data processing end can determine the first output value according to a fourth formula. The fourth formula includes: f nt =|r n |×S nt The f nt Represents the first output value; the |r n | represents the correlation coefficient; the S nt This represents the first state value.

[0108] Understandably, the data processing end can also determine the first output value based on other formulas. This application embodiment does not limit this and can be configured according to actual needs. For example, a new formula can be obtained by equivalently substituting some content in the fourth formula.

[0109] In this way, by determining the first output value using the first state value and the correlation coefficient, and taking into account the influence of the correlation coefficient between the input and output on the output value, the accuracy of the prediction results can be further improved.

[0110] The process by which the S301 data processing terminal obtains the historical output set is described below. This process may include, but is not limited to, method A or method B described below.

[0111] Method A: Use at least two historical output values ​​included in the first set of data collection as the historical output set;

[0112] Method B: Take the historical output values ​​after deleting discrete data from at least two historical output values ​​included in the first set of collection as the historical output set.

[0113] For method A, the data processing end directly uses at least two historical output values ​​from the first set of data collected as the historical output set. Example 1: The data processing end uses the collected sales data for August as the first set. The first set includes y1, y2, ..., y30; y represents daily sales, and y plus the data indicates the day of August for which the daily sales were received; for example, y1 represents the sales on August 1st; y2 represents the sales on August 2nd; and y30 represents the sales on August 30th.

[0114] Method A is simple and easy to implement.

[0115] like Figure 4 As shown, method B may include, but is not limited to, S401 to S405 below.

[0116] S401. The data processing end clusters the first set to obtain at least one cluster.

[0117] The first set includes at least two historical output values; each of the at least one cluster includes at least one historical output value.

[0118] This application does not limit the specific clustering method, and it can be configured according to actual needs. For example, a centroid can be defined for each cluster, and clustering can be performed based on the distance from each historical output value to the centroid to obtain at least one cluster. For example, clustering can be performed randomly.

[0119] Based on Example 1, S401 can be Example 2: perform initial clustering on the sales revenue of 8, dividing the sales revenue of 1 to 10, 11 to 20, and 21 to 30 into 3 clusters. That is, the initial division is y1 to y10 as cluster k1, y11 to y20 as cluster k2, and y21 to y30 as cluster k3.

[0120] S402, The data processing end performs a second processing on each of the at least one cluster to obtain the distance between each of the at least two historical output values ​​and the center point.

[0121] The second process includes: calculating the distance between each of the historical output values ​​in at least one of the historical output values ​​included in the cluster and the center point.

[0122] The embodiments of this application do not limit the specific content of the distance, and can be configured according to actual needs. For example, the distance can be a numerical difference or other distance formulas.

[0123] For example, the distance from data yn to cluster center point Jki can be calculated using the following formula (1).

[0124]

[0125] Among them, D i y represents the distance from the data yn to the cluster center Jki; n k represents a data point in the ki cluster; i Ji represents the i-th cluster; Jki represents the center point of the i-th cluster. This center point can be the average value or the median value, etc.

[0126] S403. The data processing end adjusts the at least one cluster based on the distance between each of the at least two historical output values ​​and the center point.

[0127] This application does not limit the specific adjustment method, but mainly removes historical output values ​​that are far from the center point of the cluster from the cluster and moves historical output values ​​that are close to the center point of the cluster into the cluster. The determination of distance can be based on a distance threshold, which will not be elaborated here.

[0128] S404. The data processing end determines whether the historical output values ​​included in each of the adjusted clusters have converged.

[0129] For example, the convergence of historical output values ​​can be determined based on the normal distribution. That is, if all historical output values ​​in a cluster satisfy the overall distribution, then all historical output values ​​in that cluster are considered to have converged.

[0130] For example, if a cluster contains historical output values ​​that satisfy the following formula (2), then the cluster is considered to be normally distributed.

[0131]

[0132] Where, k i Denotes the k-th cluster; exp represents the expectation; y n This represents a data point in the k-th cluster.

[0133] It should be noted that if some data in a cluster cannot converge, then that data is considered as discrete data in that cluster.

[0134] It should also be noted that the clusters in the first set can be adjusted multiple times before executing S404.

[0135] S405. The data processing end deletes the discrete data included in each cluster to obtain the historical output value.

[0136] The discrete data refers to data that cannot converge.

[0137] The data processing unit deletes the discrete data included in each cluster to obtain the historical output value.

[0138] For method B, since the discrete data in the first set is deleted, the accuracy of the range of the first set can be improved, thereby further improving the accuracy of the predicted output.

[0139] The following describes the process by which the S302 data processing terminal calculates the correlation coefficient between the input and output based on the historical output set and the various types of historical input values. This process may include, but is not limited to, Embodiment 1 or Embodiment 2 described below.

[0140] Implementation 1: Calculate the correlation coefficient based on the historical output set, which exists as a discrete first set.

[0141] Implementation 2: Calculate the correlation coefficient based on the historical output set that exists in clusters.

[0142] The process of calculating the correlation coefficient based on the historical output set, which exists as a discrete first set, is described below. This process may include, but is not limited to, S501 to S503 described below.

[0143] S501, The data processing end performs a first processing for each of the plurality of types to obtain the relevance corresponding to the plurality of types.

[0144] The first process includes: calculating the correlation between the historical input values ​​of the type and the historical output set based on the historical input values ​​of the type and the historical output set.

[0145] The embodiments of this application do not impose a unique limitation on the specific method for calculating the correlation between a type of historical input value and historical output set, and can be configured according to actual needs.

[0146] For example, the correlation between a type of historical input value and historical output set can be calculated according to the following formula (3).

[0147]

[0148] Where x represents the independent variable, i.e., the historical input value; and y represents the dependent variable, i.e., the historical output value. This represents the evaluation value of the historical input values ​​for this type; represents the evaluation value of the historical output values ​​in the historical output set; r represents the correlation between x and y, that is, the correlation between the historical input values ​​of a certain type and the historical output set.

[0149] S502. The data processing end determines the type that meets the first condition among the relevance of the multiple types as the reference type.

[0150] The first condition is used to help determine the reference type. The specific content of the first condition is not uniquely limited in the embodiments of this application, and can be configured according to actual needs.

[0151] In one possible implementation, the first condition may include: the relevance between the data and the historical output set is greater than or equal to a relevance threshold. The relevance threshold can be configured according to actual needs, and this embodiment of the application does not limit it. An exemplary relevance threshold may be 0.8.

[0152] In another possible implementation, the first condition may include: the first one or several types of historical input values ​​and historical output sets after sorting the relevance between multiple types of historical input values ​​and historical output sets from largest to smallest.

[0153] S503. The data processing end determines the correlation coefficient based on the historical input values ​​of the reference type and the historical output set.

[0154] The process by which the data processing end determines the correlation coefficient based on the historical input values ​​and historical output set of the reference type is similar to the process in S501 where the data processing end calculates the correlation between the historical input values ​​and the historical output set of the type. For details, please refer to the relevant description in S501, which will not be repeated here.

[0155] Understandably, if there is only one reference type, the relevance of the reference type can be used as the correlation coefficient; if there are multiple reference types, the evaluation values ​​of the relevance of the multiple reference types can be used as the correlation coefficient.

[0156] Compared with implementation 2, implementation 1 is simpler and easier to implement.

[0157] For implementation 2, unlike implementation 1, since the historical output set exists in the form of clusters, the correlation between the historical input value of the type and the historical output set based on the historical input value of the type and the historical output set can be implemented as follows: the data processing end calculates the correlation between the historical input value of the type and the historical output value of the cluster based on the historical input value of the type and the historical output value included in a cluster, and then takes the average of the correlation between the historical input value of the type and all clusters as the correlation between the historical input value of the type and the historical output set.

[0158] For the specific method by which the data processing end calculates the correlation between the historical input value of the type and the historical output value of the cluster based on the historical input value of the type and the historical output value of a cluster, please refer to the relevant description in S501, which will not be repeated here.

[0159] Compared with Implementation 1, the correlation coefficient determined by the formula in Implementation 2 has higher accuracy, which can further improve the accuracy of the prediction results.

[0160] The following uses a recurrent neural network as an example to illustrate the data processing method provided in the embodiments of this application.

[0161] For ease of understanding, some terms used in the following embodiments are explained.

[0162] Clustering: Dividing a dataset into different classes or clusters according to a specific criterion, so that the similarity of data objects within the same cluster is as high as possible, and the differences of data objects in different clusters are also as high as possible.

[0163] Linear dependence: In linear algebra, if a vector in a vector space cannot be represented by a linear combination of a finite number of other vectors, it is said to be linearly independent; otherwise, it is said to be linearly dependent.

[0164] Chi-square test: The degree of deviation between the actual observed values ​​and the theoretically inferred values ​​of a statistical sample.

[0165] Recurrent Neural Network: A recurrent neural network that takes sequence data as input, recursively moves along the direction of sequence evolution, and all nodes are connected in a chain-like manner.

[0166] With the rapid development of the internet industry, data analysis is playing an increasingly important role in business practices. Currently, data analysis processes involve large sample sizes, relatively complex calculations, and the data sampling dimensions are mostly determined by experienced business personnel. This leads to excessive human intervention in the data analysis results, making it difficult to establish stable correlations between input and output items. A separate regression process is also needed for the analysis results to ensure that the goodness of fit of the data analysis model is within an acceptable range.

[0167] The relevant technologies have the following problems in the data analysis process:

[0168] 1) The sampled data has many dimensions, a large amount of data, and a complex analysis process.

[0169] 2) The correlation between the analysis results is low, and human intervention is still required for decision-making.

[0170] 3) The prediction results require a separate regression testing process.

[0171] This embodiment of the application has the following characteristics:

[0172] 1) Based on the aggregated categorical data, calculate the linear correlation ratio between input and output items (equivalent to the correlation mentioned above). Following the principle of normal distribution, rather than through manual intervention, identify the independent variable that has the greatest influence on the dependent variable, thereby reducing the dimensions of data analysis.

[0173] 2) Adding a validation layer to the recurrent neural network model, compared to the traditional three-layer structure of input, state, and output, allows the data regression process to be completed during the model calculation, reducing the cost of manually regressing data and modifying and optimizing the calculation model.

[0174] 3) Recurrent neural networks eliminate the biases of historical states and the errors of the current input at the state layer, and obtain the output data with the highest linear correlation by calculating the linear correlation ratio at the output layer. Finally, they obtain the analysis results with high correlation and low error, and the decision-making behavior no longer requires manual analysis and intervention.

[0175] Regarding related technologies, on the one hand, the current selection of data analysis dimensions relies on business experience and judgment, and too many reference dimensions affect the predictive ability and stability of the data model. On the other hand, existing neural network machine learning algorithms only focus on the model's predictive ability, and a separate regression step is still needed for regression testing of the prediction results. Bias in the input data can easily lead to the model's predictions fitting the data too well.

[0176] The specific implementation process of this embodiment will be described below.

[0177] 1. Clustering data

[0178] Business interactions often generate a large amount of data. First, clustering the data (equivalent to historical output values) can reduce the scattering of the sampled data and reduce the impact of large data differences on subsequent neural network learning and analysis.

[0179] Clustering processes may include, but are not limited to, Figure 6 Steps A1 to A5 are shown.

[0180] Step A1: Divide the data into k clusters.

[0181] Here, "data" refers to the historical output set. Each data point represents a single historical output value.

[0182] Step A2: Calculate the average value ki for each cluster and use ki as the center point of that cluster.

[0183] Step A3: Calculate the distance from each data point yn to the cluster center point Jki.

[0184] In one possible implementation, the standard deviation can be used to represent the data distance.

[0185] For example, the distance from data yn to cluster center point Jki can be calculated using the following formula (1).

[0186]

[0187] Among them, D i y represents the distance from the data yn to the cluster center Jki; n k represents a data point in the ki cluster; i This represents the i-th cluster.

[0188] Step A4: Based on the data distance, re-divide the data yn into cluster kn.

[0189] Step A5: Determine whether the data follows a normal distribution.

[0190] If the data follows the overall distribution, the process ends; if the data does not follow a normal distribution, repeat steps A2 to A4 until the data converges and follows a normal distribution.

[0191] For example, if the data satisfies the following formula (2), then the data is considered to be normally distributed.

[0192]

[0193] Where, k i Denotes the k-th cluster; exp represents the expectation; y n This represents a data point in the k-th cluster.

[0194] The following example, Example A, illustrates the process of clustering data.

[0195] Suppose we have a batch of sales data for August, with daily sales figures y1, y2, ..., y30.

[0196] 1) First, perform initial clustering of the data, dividing data 1 to 10, 11 to 20, and 21 to 30 into 3 clusters. That is, initially, y1 to y10 is cluster k1, y11 to y20 is cluster k2, and y21 to y30 is cluster k3.

[0197] 2) Calculate the average value ki for each cluster and use ki as the center point of that cluster.

[0198] Specifically, calculate the average value of cluster k1, Jk1 = (y1 + y2 + ... + y10) / 10; and take the average value of cluster k1, Jk1, as the center point of the cluster.

[0199] Calculate the average value of cluster k2, Jk2 = (y11 + y12 + ... + y20) / 10; and take the average value of cluster k2, Jk2, as the center point of the cluster.

[0200] Calculate the average value of cluster k3, Jk3 = (y21 + y22 + ... + y30) / 10; and take the average value of cluster k3, Jk3, as the center point of the cluster.

[0201] 3) Calculate the distance from each data point yn to the cluster center point ki.

[0202]

[0203] Where D1 represents the distance from data y1 to the cluster center Jk1; y1 is a data point in cluster k1; Jk1 represents the center point of cluster k1; Dn represents the distance from data y30 to the cluster center Jki; y 30 Jki represents a data point in the ki cluster; Jki represents the center point of the ki cluster.

[0204] 4) Based on the data distance, re-divide the data into clusters k1, k2, and k3, and determine whether the data in each cluster satisfies a normal distribution.

[0205] The final partitioning result is:

[0206] k1=[y3, y4, y10, y11, y13, y15, y16, y24, y26, y27];

[0207] k2=[y1, y5, y6, y7, y12, y20, y21, y22, y25, y30];

[0208] k3=[y2, y8, y9, y14, y17, y18, y19, y23, y28, y29].

[0209] 2. Narrow down the dimensions of data analysis

[0210] Within each cluster, a linear correlation index is used to initially screen the data dimensions of the independent variables (equivalent to input values), further narrowing the scope of data analysis.

[0211] The choice of independent and dependent variables is determined by the specific business scenario. The linear coefficient between independent and dependent variables can be calculated using the following formula (3).

[0212]

[0213] Where x represents the independent variable, y represents the dependent variable, and r represents the correlation coefficient between x and y.

[0214] Specifically, the correlation coefficient r ranges from [-1, 1], and the linear correlation is represented by the range of values ​​for |r|.

[0215] 1) |r| < 0.3 indicates a low degree of linear correlation;

[0216] 2) 0.3 <= |r| < 0.5 indicates a low to medium degree of linear correlation;

[0217] 3) 0.5 <= |r| < 0.8 indicates moderate linear correlation;

[0218] 4) 0.8 <= |r| < 1.0 indicates a high degree of linear correlation.

[0219] To minimize the scope of data analysis, this embodiment selects only the independent variable x, which is highly linearly correlated with the dependent variable y, i.e., the independent variable with a linear correlation coefficient r of 0.8 <= |r| < 1.0 with respect to the dependent variable y.

[0220] Using the daily sales figures in Example 1 as an example, we will explain the process of narrowing down the dimensions of data analysis.

[0221] Example B: Suppose that there are multiple independent variables xn affecting the dependent variable daily sales yn, namely: daily channel advertising volume x1n, daily user pageviews x2n, daily user insurance clicks x3n, and daily user payment clicks x4n. This process of narrowing down the dimensions of data analysis can include:

[0222] 1) Calculate the mean of each independent variable:

[0223] For the independent variable x1n, the average value of the independent variable x1n is Jx1 = (x11 + x12 + ... + x130) / 30;

[0224] For the independent variable x2n, the average value of the independent variable x2n is Jx2=(x21+x22+……+x230) / 30;

[0225] For the independent variable x3n, the average value of the independent variable x3n is Jx3=(x31+x32+……+x330) / 30;

[0226] For the independent variable x4n, the average value of the independent variable x4n is Jx4=(x41+x42+……+x430) / 30.

[0227] 2) Calculate the linear phase coefficient between each independent variable and the dependent variable, daily sales revenue.

[0228] For the independent variable x1n, the linear correlation coefficient between the independent variable x1n and the dependent variable y is r1, and

[0229] For the independent variable x2n, the linear correlation coefficient between the independent variable x2n and the dependent variable y is r2, and

[0230] For the independent variable x3n, the linear correlation coefficient between the independent variable x3n and the dependent variable y is r3, and

[0231] For the independent variable x4n, the linear correlation coefficient between the independent variable x4n and the dependent variable y is r4, and

[0232] If the linear correlation coefficient r between the independent variable and the dependent variable is greater than or equal to 0.8 and less than or equal to 1, then the independent variable is determined to be a factor xn that has a significant impact on daily sales yn.

[0233] 3. Learn data patterns and use regression to predict conclusions.

[0234] The recurrent neural network designed in this embodiment of the application includes: an input layer X, a state layer S, an output layer Y, and a validation layer C. The input layer, state layer, and output layer primarily possess data learning and prediction capabilities, such as... Figure 7 As shown, the input layer X passes the input data xt to the state layer S. The state layer S calculates xt to obtain the state St, and then passes St to the output layer Y. The output layer Y outputs the predicted output result ft. Since regression validation is also a crucial step in data analysis, this embodiment of the application performs regression testing between the predicted output ft and the actual result yt in the validation layer C to obtain the deviation Ct. This enables the entire model to have self-validation capabilities, reducing the tendency to overfit the data.

[0235] To facilitate understanding, some of the symbols are explained below.

[0236] xnt represents each input; where n represents different input values, for example, x1 represents the daily ad placement volume of the channel, x2 represents the daily pageviews of users, x3 represents the daily insurance clicks of users, x4 represents the daily payment clicks of users; t represents the specific number of times. For example, t can represent the current time.

[0237] rn represents the linear correlation coefficient between the corresponding input and output;

[0238] fnt represents the output for each iteration, i.e., the predicted value for this input.

[0239] ynt represents the actual output value for each iteration;

[0240] Snt represents the state recorded by the neural network at each step;

[0241] Cnt represents the result of each data regression.

[0242] Specifically, Cnt can be calculated using the following formula (4) (equivalent to the second formula).

[0243]

[0244] Among them, C nt This represents the proportion of deviation between the actual output and the predicted output for each instance; y nt f represents the actual output value for each iteration. ntThis represents the actual output value for each time.

[0245] The state (equivalent to the first state value) of each neural network record can be calculated using the following formulas (5) (equivalent to the third formula) and (6) (equivalent to the first formula).

[0246] S nt =[(1-C nt )×S n(t-1) +F(x nt )] Formula (5);

[0247]

[0248] Among them, S nt This indicates the recording result of the current state layer; C nt Indicates the proportion of the deviation between the predicted value and the actual value; 1-C nt This indicates the reference ratio for this instance; S n(t-1) This represents the result of the previous record; F(x) nt () represents the reference value for this input; x nt This represents each input; This represents the average value of the input.

[0249] The output fnt for each iteration can be calculated using the following formula (7) (equivalent to the fourth formula).

[0250] f nt =|r n |×S nt Formula (7);

[0251] Among them, f nt This represents the actual output value; |r n | represents the linear correlation coefficient between the input and output; S nt This indicates the state of each record.

[0252] Below, we will calculate C. nt The process will be explained.

[0253] For the input value xnt, the neural network calculates the predicted output value ft according to formulas (4) to (7) above. In reality, the actual output value is yt. Since the predicted output value ft and the actual output value yt are not necessarily exactly the same, the deviation ratio Ct is used to represent the deviation ratio between the predicted output value ft and the actual output value yt.

[0254] The smaller the deviation ratio, the more accurate the prediction, the more reliable the result of this calculation, and the greater the proportion that can be used as a reference for the next calculation.

[0255] Specifically, the deviation ratio between the actual output and the predicted output can be calculated by the ratio of the deviation between the actual output value and the predicted output value to the predicted output value, which can be described by the following formula (8).

[0256]

[0257] Among them, bias measures the degree of deviation between the predicted value calculated by the neural network and the actual result, characterizing the fitting ability of the neural network model itself. "Deviation between actual output value and predicted output value" indicates the magnitude of the deviation between the actual output value and the predicted output value, that is, the deviation value between the actual output value and the predicted output value; "Deviation between actual output value and predicted output value / predicted output value" indicates the proportion of the difference value in the prediction result, that is, the proportion of the deviation between the predicted value and the actual value.

[0258] For example, Cnt can be calculated using the above formula (4).

[0259]

[0260] Among them, C nt This represents the proportion of deviation between the actual output and the predicted output for each instance; y nt f represents the actual output value for each iteration. nt This represents the actual output value for each time.

[0261] The calculation process of St is explained below.

[0262] The state recorded by the neural network is a preservation of the results of the current calculation. The recurrent neural network traces the history through the state layer St, so that each output is not only related to the current input, but also to the historical input, thus having the learning ability to "discover historical patterns".

[0263] For the input value xnt, F(x) can be calculated using the following formula (9). nt ).

[0264]

[0265] Variance measures the change in neural network learning performance caused by variations in the same size dataset, characterizing the impact of input data perturbations. "Variance between input value and input mean" represents the possible error value in the current input data; "Variance between input value and input mean / input value" represents the error percentage in the current input; (1 – error percentage) × input value represents the data worth referencing in the current input.

[0266] Since the calculation result obtained from the input value is only a part of the state layer parameters, the learning ability of a recurrent neural network is mainly reflected in its ability to infer the current output result from the past. Formulaically, this means that the previous state influences the current state. As mentioned above, the calculation of a neural network has a bias; the deviation ratio between the predicted value and the actual value is Cnt. Therefore, the reference ratio of the previous state is (1-Cnt). Thus, the current state includes a portion of (1-Cnt) × the previous state, i.e., (1-Ct) × S. n(t-1) .

[0267] The final state layer can be obtained using the following formula (10). nt =[(1-output deviation ratio)×previous state value+input calculation result] formula (10).

[0268] For example, the state of each neural network record can be calculated using the above formulas (5) and (6).

[0269] S nt =[(1-C nt )×S n(t-1) +F(x nt )] Formula (5);

[0270]

[0271] Among them, S nt This indicates the recording result of the current state layer; C nt Indicates the proportion of the deviation between the predicted value and the actual value; 1-C nt This indicates the reference ratio for this instance; S n(t-1) This represents the result of the previous record; F(x) nt () represents the reference value for this input; x nt This represents each input; This represents the average value of the input.

[0272] Below, on f nt The calculation process will be explained.

[0273] Since the results recorded by the state layer are not the final output, and each input value and output have a linear correlation coefficient rn, for the state value, only the part whose linear correlation coefficient with the output item satisfies the condition is taken to obtain the final output result, that is, the output prediction value.

[0274] Specifically, it can be obtained through the following formula (11).

[0275] f nt =|linear correlation coefficient| × current state value formula (11);

[0276] For example, the output fnt for each time can be calculated using the following formula (7).

[0277] f nt =|r n |×S nt Formula (7);

[0278] Among them, f nt This represents the actual output value; |r n | represents the linear correlation coefficient between the input and output; S nt This indicates the state of each record.

[0279] It is understandable that the output prediction value calculated by formulas (8) to (11) above can be regarded as the result with the highest linear correlation with the output item and the smallest error.

[0280] Based on examples A and B above, the conclusion of learning data patterns and regression prediction can be represented as example C:

[0281] The neural network only has the input value xn1 during the first processing, without the state value from the previous calculation. Therefore, the first S... n1 It can be described as That is, the first obtained state dataset [S11, S21, ..., Sn1] = [(1-(x11-x1)] 2 / (x11)×x11,(1-(x21-x1)) 2 / x21)×x21,……,(1-(xn1-x1) 2 / xn1)×xn1].

[0282] Then, the predicted value dataset for the first output is calculated using the linear correlation coefficient.

[0283] [f11, f21,..., fn1]=[|r1|×S11, |r2|×S21,..., |rn|×Sn1].

[0284] Starting from the second processing step, since the actual output value can already be collected, and the neural network has already recorded the state from the first processing step, each subsequent input can go through the complete computation process:

[0285] Calculate the deviation ratio Cnt = (ynt - fnt) 2 / fnt;

[0286] The deviation ratio dataset [C1t, C2t, ..., Cnt] = [(y1t - f1t)] can be obtained. 2 / f1t, (y2t-f2t) 2 / f2t, ..., (ynt-fnt) 2 / fnt];

[0287] Calculate the state: Snt = [(1-Ct)×Sn(t-1)+(1-(xnt-xt)] 2 / xnt)×xnt];

[0288] The state dataset [S1t, S2t, ..., Snt] can be obtained as [[(1-Ct)×S1(t-1)+(1-(x1t-xt)]]. 2 / x1t)×x1t], [(1-Ct)×S2(t-1)+(1-(x2t-xt) 2 / x2t)×x2t],……,[(1-Ct)×Sn(t-1)+(1-(xnt-xt) 2 / xnt)×xnt]];

[0289] Calculate the predicted value: fnt = |rn| × Sn;

[0290] The final output dataset [f1t, f2t, ..., fnt] = [|r1|×S1t, |r2|×S2t, ..., |rn|×Snt].

[0291] This embodiment of the application has the following technical effects:

[0292] First, by using aggregation algorithms, the dispersion of data is reduced, making the connections between data points closer and facilitating the analysis of patterns and relationships between them.

[0293] Secondly, by calculating linear correlation, we can analyze the linear correlation ratio (equivalent to the correlation coefficient) between the input and output items. This can reduce the dimensionality of data analysis and simplify the data scope. Furthermore, in recurrent neural network calculations, the linear correlation coefficient can be used to obtain the prediction result that is most correlated with the output item.

[0294] Furthermore, in the computational process of a recurrent neural network, the deviation between the predicted and actual values ​​is calculated on the one hand, and the deviation values ​​from historical states and the error values ​​of the current input are eliminated on the other hand. Therefore, the results of the recurrent neural network calculation tend to be more stable and accurate. This eliminates the trouble of manual intervention in data regression, and the model calculation results will not have the problem of overfitting.

[0295] To implement the above data processing method, an embodiment of this application provides a data processing apparatus, which is described below in conjunction with... Figure 8 The structural diagram of the data processing device shown is used for illustration.

[0296] like Figure 8As shown, the data processing device 80 includes: a first determining unit 801, a second determining unit 802, a third determining unit 803, and a fourth determining unit 804. Wherein:

[0297] The first determining unit 801 is used to determine an input reference value based on a first input value; the input reference value is used to characterize the magnitude of the first output value that affects the first input value.

[0298] The second determining unit 802 is used to determine the deviation degree based on the second output value and the second actual value; the deviation degree is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value;

[0299] The third determining unit 803 is used to determine a first state value based on the input reference value, the deviation, and the second state value; the second state value is the state value corresponding to the second input value.

[0300] The fourth determining unit 804 is used to determine a first output value based at least on the first state value.

[0301] In some embodiments, the first determining unit 801 is further configured to:

[0302] The input reference value is determined according to the first formula;

[0303] The first formula includes: The F(x) nt ) represents the input reference value; the x nt This represents the first input value; the This represents the average value of the input.

[0304] In some embodiments, the second determining unit 802 is further configured to:

[0305] The deviation is determined according to the second formula;

[0306] The second formula includes: The C nt Indicates the deviation; the y nt This represents the second actual value; the f nt ' indicates the second output value.

[0307] In some embodiments, the third determining unit 803 is further configured to:

[0308] The first state value is determined according to the third formula;

[0309] The third formula includes: S nt =[(1-C nt )×S n(t-1) +F(x nt )]; The S nt This represents the first state value; the C nt The deviation is represented by F(x). nt ) represents the input reference value, and the S n(t-1) This represents the second state value.

[0310] In some embodiments, the fourth determining unit 804 is further configured to:

[0311] Obtain the historical output set, as well as historical input values ​​of multiple types;

[0312] Based on the historical output set and the historical input values ​​of the multiple types, calculate the correlation coefficient between the input and the output;

[0313] The first output value is determined according to the fourth formula;

[0314] The fourth formula includes: f nt =|r n |×S nt The f nt Represents the first output value; the |r n | represents the correlation coefficient; the S nt This represents the first state value.

[0315] In some embodiments, the fourth determining unit 804 is further configured to:

[0316] Clustering the first set yields at least one cluster; the first set includes at least two historical output values; each of the at least one cluster includes at least one historical output value.

[0317] For each of the at least one cluster, a second process is performed to obtain the distance between each of the at least two historical output values ​​and the center point; the second process includes: calculating the distance between each of the at least one historical output value included in the cluster and the center point;

[0318] The at least one cluster is adjusted based on the distance between each of the at least two historical output values ​​and the center point;

[0319] Determine whether the adjusted historical output values ​​of each cluster have converged;

[0320] The discrete data included in each of the clusters is deleted to obtain the historical output set; the discrete data is data that cannot converge.

[0321] In some embodiments, the fourth determining unit 804 is further configured to:

[0322] For each of the plurality of types, a first process is performed to obtain the relevance corresponding to the plurality of types; wherein, the first process includes: calculating the relevance between the historical input values ​​of the type and the historical output set based on the historical input values ​​of the type and the historical output set;

[0323] Among the relevance values ​​corresponding to the multiple types, the type that meets the first condition is determined as the reference type;

[0324] The correlation coefficient is determined based on the historical input values ​​of the reference type and the historical output set. It should be noted that the data processing apparatus provided in this application embodiment includes all the units included, which can be implemented by a processor in an electronic device; of course, it can also be implemented by specific logic circuits; in implementation, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP), or a field-programmable gate array (FPGA), etc.

[0325] The descriptions of the above device embodiments are similar to those of the above method embodiments, and have similar beneficial effects. For technical details not disclosed in the device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0326] It should be noted that, in the embodiments of this application, if the above-described data processing method is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, or the part that contributes to the related technology, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), magnetic disks, or optical disks. Thus, the embodiments of this application are not limited to any specific hardware and software combination.

[0327] To implement the above data processing method, this application provides an electronic device, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the steps in the data processing method provided in the above embodiments.

[0328] The following is combined Figure 9 The electronic device 90 shown is illustrated with a structural diagram of the electronic device.

[0329] In one example, electronic device 90 can be the aforementioned electronic device. For example... Figure 9 As shown, the electronic device 90 includes: a processor 901, at least one communication bus 902, a user interface 903, at least one external communication interface 904, and a memory 905. The communication bus 902 is configured to enable communication between these components. The user interface 903 may include a display screen, and the external communication interface 904 may include standard wired and wireless interfaces.

[0330] The memory 905 is configured to store instructions and applications executable by the processor 901, and can also cache data to be processed or already processed by the processor 901 and various modules in the electronic device (e.g., image data, audio data, voice communication data and video communication data), which can be implemented by flash memory or random access memory (RAM).

[0331] Fourthly, embodiments of this application provide a storage medium, namely a computer-readable storage medium, on which a computer program is stored, which, when executed by a processor, implements the steps in the data processing method provided in the above embodiments.

[0332] It should be noted that the descriptions of the storage medium and device embodiments above are similar to the descriptions of the method embodiments above, and have similar beneficial effects. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the descriptions of the method embodiments of this application for understanding.

[0333] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "in one embodiment" or "in some embodiments" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. It should be understood that in the various embodiments of this application, the sequence numbers of the above-described processes do not imply a sequential order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. The sequence numbers of the above-described embodiments are merely descriptive and do not represent the superiority or inferiority of the embodiments.

[0334] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0335] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods, such as: multiple units or components can be combined, or integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the various components shown or discussed can be through some interfaces, and the indirect coupling or communication connection between devices or units can be electrical, mechanical, or other forms.

[0336] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units. They may be located in one place or distributed across multiple network units. Some or all of the units may be selected to achieve the purpose of this embodiment according to actual needs.

[0337] In addition, each functional unit in the various embodiments of this application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be implemented in hardware or in the form of hardware plus software functional units.

[0338] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as mobile storage devices, read-only memory (ROM), magnetic disks, or optical disks.

[0339] Alternatively, if the integrated units described above are implemented as software functional modules and sold or used as independent products, they can also be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, or the parts that contribute to related technologies, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as mobile storage devices, ROMs, magnetic disks, or optical disks.

[0340] The above description is merely an embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A data processing method, characterized in that, The method includes: Based on the first input value, an input reference value is determined; the input reference value is used to characterize the magnitude of the first output value that affects the first input value. Based on the second output value and the second actual value, the deviation is determined; the deviation is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value; Based on the input reference value, the deviation, and the second state value, a first state value is determined; the second state value is the state value corresponding to the second input value. Obtain the historical output set, as well as historical input values ​​of multiple types; Based on the historical output set and the historical input values ​​of the multiple types, calculate the correlation coefficient between the input and the output; The first output value is determined according to the fourth formula; The fourth formula includes: The This represents the first output value; the The correlation coefficient represents the correlation coefficient; This represents the first state value.

2. The method according to claim 1, characterized in that, The step of determining the input reference value based on the first input value includes: The input reference value is determined according to the first formula; The first formula includes: The The input reference value is represented; This represents the first input value; the This represents the average value of the input.

3. The method according to claim 1, characterized in that, The determination of deviation based on the second output value and the second actual value includes: The deviation is determined according to the second formula; The second formula includes: The Indicates the degree of deviation; This represents the second actual value; the This represents the second output value.

4. The method according to claim 1, characterized in that, Determining the first state value based on the input reference value, the deviation, and the second state value includes: The first state value is determined according to the third formula; The third formula includes: The This represents the first state value; the Indicates the degree of deviation, the This represents the input reference value, the This represents the second state value.

5. The method according to claim 1, characterized in that, The obtained historical output set includes: Clustering the first set yields at least one cluster; the first set includes at least two historical output values; each of the at least one cluster includes at least one historical output value. For each of the at least one cluster, a second process is performed to obtain the distance between each of the at least two historical output values ​​and the center point; the second process includes: calculating the distance between each of the at least one historical output value included in the cluster and the center point; The at least one cluster is adjusted based on the distance between each of the at least two historical output values ​​and the center point; Determine whether the adjusted historical output values ​​of each cluster have converged; The discrete data included in each of the clusters is deleted to obtain the historical output set; the discrete data is data that cannot converge.

6. The method according to claim 1, characterized in that, The step of calculating the correlation coefficient between input and output based on the historical output set and the multiple types of historical input values ​​includes: For each of the plurality of types, a first process is performed to obtain the relevance corresponding to the plurality of types; wherein, the first process includes: calculating the relevance between the historical input values ​​of the type and the historical output set based on the historical input values ​​of the type and the historical output set; Among the relevance values ​​corresponding to the multiple types, the type that meets the first condition is determined as the reference type; The correlation coefficient is determined based on the historical input values ​​of the reference type and the historical output set.

7. A data processing apparatus, characterized in that, The device includes: The first determining unit is configured to determine an input reference value based on a first input value; the input reference value is used to characterize the magnitude of the first output value that affects the first input value. The second determining unit is used to determine the deviation degree based on the second output value and the second actual value; the deviation degree is used to characterize the degree of deviation between the second output value and the second actual value; the second output value is the output value corresponding to the second input value, the second actual value is the actual value corresponding to the second input value, and the second input value is a historical input value; The third determining unit is used to determine a first state value based on the input reference value, the deviation, and the second state value; the second state value is the state value corresponding to the second input value. The fourth determining unit is used to determine a first output value based at least on the first state value; The fourth determining unit is also used for: Obtain the historical output set, as well as historical input values ​​of multiple types; Based on the historical output set and the historical input values ​​of the multiple types, calculate the correlation coefficient between the input and the output; The first output value is determined according to the fourth formula; The fourth formula includes: The This represents the first output value; the The correlation coefficient represents the correlation coefficient; This represents the first state value.

8. An electronic device comprising a memory and a processor, the memory storing a computer program executable on the processor, the processor executing the program to implement the data processing method according to any one of claims 1 to 6.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the data processing method according to any one of claims 1 to 6.

Images