Data recording and analysis method, device, equipment and storage medium

Abstract

The present application relates to a data record analysis method, apparatus, device and storage medium, the method comprising: obtaining a data record storage address in response to a data record analysis instruction; accessing the data record storage address to obtain a data record set, wherein each data record in the data record includes a first indicator and a second indicator; randomly assigning each data record in the data record set to a data bucket; determining the first indicator, the second indicator and the number of data records as indicators to be analyzed; obtaining a first analysis target and obtaining a second analysis target; determining the covariance between the first analysis target and the second analysis target based on the bucket average indicator total value; and determining the analysis result of the data record based on the covariance between the first analysis target and the second analysis target. The present application performs dimensionality reduction processing on the data records and utilizes the bucket average indicator total value to quickly obtain the analysis result of the data record, thereby improving the efficiency of data record analysis.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a data recording and analysis method, apparatus, device, and storage medium. Background Art

[0002] AB testing is a method for optimizing products by collecting and analyzing user feedback data. It can be applied to a wide range of business scenarios, including user experience optimization, conversion rate optimization, and advertising optimization. When analyzing test data from AB testing, the statistical processing method of covariance estimation is often used to assess the impact of multiple correlated or independent variables to obtain analytical results for the test data. However, covariance estimation requires significant performance overhead when the number of data records, such as test data, is large. This limits its application in large data volumes, making it an urgent challenge to quickly obtain analytical results using covariance estimation. Summary of the Invention

[0003] The present application provides a data record analysis method, device, equipment and storage medium. The present application performs dimensionality reduction processing on the data records and uses the total value of the bucket average index to quickly obtain the analysis results of the data records, thereby improving the efficiency of data record analysis.

[0004] In one aspect, the present application provides a data record analysis method, the method comprising:

[0005] In response to the data record analysis instruction, obtaining a data record storage address;

[0006] Accessing the data record storage address to obtain a data record set, wherein each data record in the data record set includes a first indicator and a second indicator;

[0007] Randomly assigning each data record in the data record set to a data bucket;

[0008] Determining the first indicator, the second indicator, and the number of data records as indicators to be analyzed;

[0009] Obtaining a first analysis objective, the first analysis objective being used to analyze a first data relationship between bucket-averaged total index values of at least two of the indicators to be analyzed; and obtaining a second analysis objective, the second analysis objective being used to analyze a second data relationship between bucket-averaged total index values of at least two of the indicators to be analyzed;

[0010] Determine the covariance between the first analysis target and the second analysis target according to the bucket average indicator total value;

[0011] An analysis result of the data record is determined based on the covariance between the first analysis target and the second analysis target.

[0012] Another aspect provides a data recording and analysis device, the device comprising:

[0013] a storage address acquisition module, configured to acquire a data record storage address in response to a data record analysis instruction;

[0014] a data record determination module, configured to access the data record storage address to obtain a data record set, wherein each data record in the data record set includes a first indicator and a second indicator;

[0015] A bucket processing module, configured to randomly assign each data record in the data record set to a data bucket;

[0016] an analysis indicator determination module, configured to determine the first indicator, the second indicator, and the number of data records as indicators to be analyzed;

[0017] An analysis target acquisition module is configured to acquire a first analysis target, the first analysis target being used to analyze a first data relationship between bucket-averaged indicator total values of at least two of the indicators to be analyzed; and to acquire a second analysis target, the second analysis target being used to analyze a second data relationship between bucket-averaged indicator total values of at least two of the indicators to be analyzed;

[0018] a covariance determination module, configured to determine the covariance between the first analysis target and the second analysis target according to the bucket average index total value;

[0019] An analysis result determination module is used to determine an analysis result of the data record according to the covariance between the first analysis target and the second analysis target.

[0020] On the other hand, an electronic device is provided, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded by the processor and executes the data recording and analysis method as described above.

[0021] On the other hand, a computer-readable storage medium is provided, in which at least one instruction or at least one program is stored. The at least one instruction or the at least one program is loaded and executed by a processor to implement the data record analysis method as described above.

[0022] This application uses a bucketing method to randomly assign each data record in a data record set to a data bucket, thereby achieving the effect of reducing the dimensionality of the original data record set. When calculating the covariance between the first analysis target and the second analysis target, the total value of the bucket average indicator can be directly used for analysis, thereby improving the calculation and storage rate of data record analysis, so that the analysis results of the data records can be obtained quickly, and solving the problem of using covariance estimation to quickly obtain the analysis results of data records in large data volume scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or the prior art, the following is a brief introduction to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0024] Figure 1 This is a schematic diagram of the implementation environment of a data recording and analysis method provided in an embodiment of the present application.

[0025] Figure 2 This is a flow chart of a data recording and analysis method provided in an embodiment of the present application.

[0026] Figure 3 This is an example diagram of random distribution of data records provided in an embodiment of the present application.

[0027] Figure 4 This is a flowchart of determining the covariance between the first analysis target and the second analysis target provided in an embodiment of the present application.

[0028] Figure 5 This is a flowchart of determining the analysis object provided in an embodiment of the present application.

[0029] Figure 6 This is a flowchart of determining the third analysis target provided in an embodiment of the present application.

[0030] Figure 7 This is a flow chart of calculating the covariance between the bucket-averaged total values of two indicators to be analyzed in the analysis object provided in an embodiment of the present application.

[0031] Figure 8 This is a flow chart of another data recording and analysis method provided in an embodiment of the present application.

[0032] Figure 9 This is a flowchart of determining a data record set provided in an embodiment of the present application.

[0033] Figure 10It is a structural diagram of a data recording and analysis device provided in an embodiment of the present application.

[0034] Figure 11 This is a schematic diagram of the hardware structure of a device provided in an embodiment of the present application for implementing the method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0035] To make the objectives, technical solutions, and advantages of this application more clear, this application will be further described in detail below with reference to the accompanying drawings. It is clear that the embodiments described are only some of the embodiments of this application, and not all of them. All other embodiments obtained by persons of ordinary skill in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0036] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0037] AB testing is a method for optimizing products by collecting and analyzing user feedback data. It primarily involves setting up two experimental groups, Group A and Group B, that differ in a parameter to be optimized (for example, the product icon color for Group A is green, while the icon color for Group B is red). Product users are randomly assigned to either Group A or Group B. The data generated by user behavior is recorded and analyzed accordingly. The results of this analysis determine which group, Group A or Group B, uses the best parameters, ultimately optimizing the product.

[0038] Covariance is primarily used to measure the degree of joint variation between two statistics. Covariance estimation can be applied in a variety of scenarios. For example, when comparing A / B test data, estimating the variance of the indicator itself is one application scenario for covariance estimation. When using historical data to reduce the variance of indicators, it is also necessary to estimate the covariance between indicators, which is another application scenario for covariance estimation. In addition, covariance estimation is also indispensable for indicator correlation analysis.

[0039] Covariance estimation has a wide range of applications, including analysis of test data and other data records. However, conventional covariance estimation methods suffer from high storage and computational overhead when the number of data records is very large, resulting in very poor analysis performance and limiting their practical application. To rapidly utilize covariance estimation to obtain analysis results for data records in large data scenarios, this application provides a data record analysis method.

[0040] See also Figure 1 , which shows a schematic diagram of an implementation environment of a data record analysis method provided by an embodiment of the present application, such as Figure 1 As shown, the implementation environment may include at least a client 01 and a server 02 .

[0041] Client 01 can be a device such as a smartphone, desktop computer, tablet computer, laptop computer, digital assistant, smart wearable device, monitoring device, or voice interaction device. It can also be software running on the device, such as a web page provided by a service provider to a user, or an application provided by the service provider to a user. Specifically, Client 01 can be used to send data record analysis instructions to Server 02 for analyzing data records.

[0042] Server 02 can be a standalone server, a distributed server, or a server cluster consisting of multiple servers. Server 02 may include a network communication unit, a processor, and a memory. Specifically, server 02 may be configured to respond to a data record analysis instruction from client 01 for analyzing data records, perform covariance estimation on the data records, and obtain analysis results for the data records.

[0043] For ease of description, a data recording and analysis method provided by an embodiment of the present application is introduced below with a server as the execution subject. This specification provides method operation steps as described in the embodiments or flow charts, but more or fewer operation steps may be included based on conventional or non-creative labor. The order of steps listed in the embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the actual system or server product is executed, it can be executed in sequence or in parallel according to the method shown in the embodiment or the accompanying drawings (for example, a parallel processor or a multi-threaded processing environment).

[0044] See also Figure 2 , which shows a flow chart of a data record analysis method provided by an embodiment of the present application. Figure 2 As shown, the method may include:

[0045] S201 , in response to a data record analysis instruction, obtaining a data record storage address.

[0046] The data record analysis instruction is used to analyze the data record, and the data record storage address represents the storage address of the data record. The data record can be any data that needs to be analyzed, such as test data, experimental data, or historical user data.

[0047] S203 , accessing the data record storage address to obtain a data record set, where each data record in the data record set includes a first indicator and a second indicator.

[0048] In actual applications, the data record storage address may store a large number of data records. For example, in the context of experimental system X, the number of data records may reach tens of millions to billions. The server may determine all data records in the data record storage address as a data record set, or may select a preset number of data records from all data records as the data record set. The preset number of data records may be set based on the specific application scenario and is not specifically limited in this application.

[0049] In the embodiments of the present application, each data record includes at least two indicators, namely a first indicator and a second indicator. The first indicator and the second indicator can be the same or different. Taking height and weight as an example, in some data record analysis scenarios, the first indicator and the second indicator can both be height or weight; in other data record analysis scenarios, the first indicator can be height and the second indicator can be weight.

[0050] S205: Randomly assign each data record in the data record set to a data bucket.

[0051] The embodiment of the present application is based on the bucketing method, which randomly assigns each data record to a data bucket, so that each data record only appears in one data bucket. Due to the random assignment, each data bucket may not contain data records, and the number of data records in each data bucket may not be equal. Figure 3 As shown in , it is an example diagram of random distribution of data records. Figure 3 Assume that there are N data records in total. If B represents the number of data buckets, H represents the first index, W represents the second index, and H represents the total number of data buckets, then i Represents the first index in the i-th data record, using W i represents the second index in the i-th data record, then the i-th data record R i It can be expressed as {H i ,W i}, then {H i ,W i} will be randomly assigned to any one of the B data buckets.

[0052] As you can understand, the larger B is, the more accurate the analysis results will be. However, setting B too large will also increase computational overhead. Typically, in the context of an experimental system like X, B can be set in the range of 50-200.

[0053] S207: Determine the first indicator, the second indicator, and the number of data records as indicators to be analyzed.

[0054] The number of data records represents the total number of data records in the data record set.

[0055] S209, obtaining a first analysis target, which is used to analyze a first data relationship between the bucket-averaged indicator total values of at least two indicators to be analyzed; and obtaining a second analysis target, which is used to analyze a second data relationship between the bucket-averaged indicator total values of at least two indicators to be analyzed.

[0056] In the embodiment of the present application, the bucket-averaged total index value of the indicator to be analyzed represents the result obtained by dividing the sum of the index values of the indicator to be analyzed in the data record set by the total number of data buckets. In one example, if the first indicator is weight, the sum of all weight values in the data record set is 1000 kg, and there are 50 data buckets in total, then the bucket-averaged total index value of weight is 1000 kg / 50 = 20 kg.

[0057] The first data relationship and the second data relationship are equivalent to a mapping. The first analysis target can be obtained by transforming the first data relationship based on the bucket-averaged total values of at least two indicators to be analyzed, and the second analysis target can be obtained by transforming the second data relationship based on the bucket-averaged total values of at least two indicators to be analyzed. In other words, the first analysis target and the second analysis target are both variables. The first analysis target is a function of the bucket-averaged total values of the at least two indicators to be analyzed corresponding to the first data relationship, and the second analysis target is a function of the bucket-averaged total values of the at least two indicators to be analyzed corresponding to the second data relationship.

[0058] It can be understood that the first analysis target is used to analyze the first data relationship between the bucket-average indicator total values of at least two of the indicators to be analyzed, and the bucket-average indicator total value can be obtained through the indicator total value of the indicator to be analyzed and the total number of data buckets. Therefore, the first analysis target is also equivalent to analyzing the first data relationship between at least two of the indicators to be analyzed.

[0059] In an example, the first indicator is height, the second indicator is weight, and the total number of data buckets is B. The bucket-averaged indicator value of the first indicator is ∑height / B, and the bucket-averaged indicator value of the second indicator is ∑weight / B. Obviously, the first data relationship between ∑height / B and ∑weight / B is equivalent to the first data relationship between height and weight.

[0060] Based on the same reason as the first analysis goal, the second analysis goal is also equivalent to analyzing the second data relationship between at least two of the indicators to be analyzed.

[0061] A function is a mapping between dependent variables and independent variables. Therefore, for any differentiable first analysis function and second analysis function, if the first analysis function is a function with the first analysis target as the independent variable and the bucket-average total index value of at least two indicators to be analyzed corresponding to the first data relationship as the dependent variable, then the first analysis target can be represented by the first analysis function; similarly, if the second analysis function is a function with the second analysis target as the independent variable and the bucket-average total index value of at least two indicators to be analyzed corresponding to the second data relationship as the dependent variable, then the second analysis target can be represented by the second analysis function.

[0062] For example, assuming that the at least two indicators to be analyzed corresponding to the first data relationship are the first indicator and the number of data records, and the at least two indicators to be analyzed corresponding to the second data relationship are the second indicator and the number of data records, and x represents the bucket-averaged total indicator value of the first indicator, y represents the bucket-averaged total indicator value of the second indicator, and z represents the bucket-averaged total indicator value of the number of data records. If the first analysis function h1 is a function with x and z as dependent variables and the first analysis target f1 as the independent variable, then h1 can be used to represent f1, that is, f1 = h1(x, z); similarly, if the second analysis function h2 is a function with y and z as dependent variables and the second analysis target f1 as the independent variable, then h2 can be used to represent f2, that is, f2 = h2(y, z).

[0063] In one example, x represents the bucket-averaged total value of the first indicator, y represents the bucket-averaged total value of the second indicator, and z represents the bucket-averaged total value of the number of data records. Assuming that the first analysis function h1(x,z)=x / z and the second analysis function h2(y,z)=y / z, if the first indicator represents height and the second indicator represents weight, the first analysis target f1 represents average height and the second analysis target f2 represents average weight.

[0064] S211, determining the covariance between the first analysis target and the second analysis target according to the bucket average indicator total value.

[0065] Please refer to the following for details: Figure 4 , which shows a flow chart of determining the covariance between the first analysis target and the second analysis target provided by an embodiment of the present application. Figure 4 As shown, step S211 may include:

[0066] S2111: Determine an analysis object according to the first analysis target and the second analysis target.

[0067] The analysis object is used to calculate the covariance between the first analysis target and the second analysis target. The analysis object determined by the server may include one or more analysis objects. The covariance between the first analysis target and the second analysis target can be determined by performing calculations based on the operational relationships between the analysis objects. In one example, A represents the covariance between the first analysis target and the second analysis target, and A = B1 + B2 - B3 + B4. The analysis objects can be B1, B2, B3, and B4. Once the values of B1, B2, B3, and B4 are determined, the value of A is also determined.

[0068] The server can directly perform a covariance operation on the first analysis target and the second analysis target to determine the analysis object. When performing the covariance operation on the first analysis target and the second analysis target, the first analysis target and the second analysis target can be respectively reduced in dimension using a vector gradient operator before performing the operation, thereby simplifying the covariance operation process.

[0069] In one possible implementation, Figure 5 As shown, step S2111 may include:

[0070] S21111, perform gradient operation and difference operation on the first analysis target to obtain a third analysis target; and perform gradient operation and difference operation on the second analysis target to obtain a fourth analysis target.

[0071] The server can use the Delta Method described below to perform gradient calculations and difference calculations based on the bucket-averaged total values of each indicator to be analyzed in the first analysis target to obtain a third analysis target; and perform gradient original acid and difference calculations based on the bucket-averaged total values of each indicator to be analyzed in the second analysis target to obtain a fourth analysis target.

[0072] The main process of using the difference method in the embodiment of the present application is as follows: if theta' represents a vector consisting of the total values of the bucket averages of at least two indicators to be analyzed, and theta represents the expectation of theta', according to the central limit theorem, when the number of data records N tends to infinity, for any differentiable function h, Obeys normal distribution. Then, according to the Delta Method, we can get:

[0073]

[0074] in, is the gradient vector of h(theta).

[0075] For the first analysis target f1 and the second analysis target f2, if f1 = h1(theta1′) and f2 = h2(theta2′), then the covariance Cov(f1,f2) between f1 and f2 can be transformed as follows:

[0076]

[0077] Based on the above utilization of the difference method, in one possible implementation, as Figure 6 As shown, in step S21111, performing gradient operation and difference operation on the first analysis target to obtain the third analysis target may include:

[0078] S601, obtaining each to-be-analyzed indicator corresponding to the first data relationship, and determining a first indicator vector according to the bucket-average indicator total value of each to-be-analyzed indicator corresponding to the first data relationship;

[0079] S603, determining the expectation of the first indicator vector as the first mean vector;

[0080] S605, calculating a partial derivative of the first analysis target with respect to the first index vector to obtain a first gradient vector;

[0081] S607, calculating the difference between the first index vector and the first mean vector to obtain a first difference vector;

[0082] S609, determining the expectation of the first gradient vector as a first expectation vector;

[0083] S611: Determine the product of the transpose of the first expected vector and the first difference vector as a third analysis target.

[0084] In an example, assuming that the indicators to be analyzed corresponding to the first data relationship are the first indicator and the number of data records, if x represents the bucket-averaged total value of the first indicator, z represents the bucket-averaged total value of the number of data records, ex represents the expectation of x, and ez represents the expectation of z, the first analysis target f1=h1(theta1′)=x / z, then the first indicator vector theta1′ can be expressed as (x,z), and the first mean vector theta1 can be expressed as (ex,ez).

[0085] In step S605, the partial derivative of the first analysis target f1 with respect to the first index vector is obtained, that is, the partial derivative of the first analysis target f1 with respect to x and z is obtained, and the obtained first gradient vector can be expressed as So that the expectation of the first gradient vector is the first expectation vector It can be expressed as Then, the third analysis target f3 obtained by performing gradient operation and difference operation on the first analysis target can be expressed as:

[0086]

[0087] Since ex and ez are fixed quantities, the content of h1(theta1) can be ignored, that is, Then, the third analysis target f3 obtained by performing gradient operation and difference operation on the first analysis target can be expressed as:

[0088]

[0089] Similarly, based on the above utilization of the difference method, performing gradient operation and difference operation on the second analysis target to obtain the fourth analysis target may include: obtaining each indicator to be analyzed corresponding to the second data relationship, and determining the second indicator vector based on the bucket average indicator total value of each indicator to be analyzed corresponding to the second data relationship; determining the expectation of the second indicator vector as the second mean vector; taking the partial derivative of the second analysis target with respect to the second indicator vector to obtain the second gradient vector; calculating the difference between the second indicator vector and the second mean vector to obtain the second difference vector; determining the expectation of the second gradient vector as the second expectation vector; and determining the product of the transpose of the second expectation vector and the second difference vector as the fourth analysis target.

[0090] In an example, assuming that the indicators to be analyzed corresponding to the second data relationship are the second indicator and the number of data records, if y represents the total bucket average indicator value of the second indicator, ey represents the expectation of y, and the second analysis target f2=h2(theta2′)=y / z, then the second indicator vector theta2′ can be expressed as (y,z), and the second mean vector theta2 can be expressed as (ey,ez).

[0091] The second gradient vector obtained by taking the partial derivative of the second analysis target f2 with respect to the second index vector theta2′, that is, taking the partial derivative of the second analysis target f2 with respect to y and z, can be expressed as So that the expectation of the second gradient vector is the second expectation vector It can be expressed as Then, the fourth analysis target f4 obtained by performing gradient operation and difference operation on the second analysis target can be expressed as:

[0092]

[0093] Since ey and ez are fixed quantities, the content of h2 (theta2) can be ignored, that is, Then, the fourth analysis target f4 obtained by performing gradient operation and difference operation on the second analysis target can also be expressed as:

[0094]

[0095] S21113, determining the analysis object according to the third analysis objective and the fourth analysis objective.

[0096] After obtaining the third analysis target and the fourth analysis target, the server can approximate the covariance between the first analysis target and the second analysis target using the covariance between the third analysis target and the fourth analysis target, and then use the covariance between the third analysis target and the fourth analysis target to determine the analysis object.

[0097] In one example, taking the example in step S21111 as an example, the covariance Cov(f1, f2) between the first analysis target f1 and the second analysis target f2 may be approximately equal to the covariance Cov(f3, f4) between the third analysis target f3 and the fourth analysis target f4, that is:

[0098]

[0099] Among them, x represents the bucket average indicator total value of the first indicator, z represents the bucket average indicator total value of the number of data records, y represents the bucket average indicator total value of the second indicator, ex represents the expectation of x, ez represents the expectation of z, and ey represents the expectation of y.

[0100] When the number of data records is large, x≈ex, y≈ey, z≈ez, and the above formula can be expressed as:

[0101]

[0102] Accordingly, the analysis object determined according to the first analysis target and the second analysis target can be as well as

[0103] S2113, calculating the covariance between the bucket average indicator total values of two indicators to be analyzed in the analysis object.

[0104] In one example, if the analysis object is The covariance between the bucket-averaged total values of the two indicators to be analyzed in the analysis object is Cov(x,y), that is, the covariance between the bucket-averaged total value x of the first indicator and the bucket-averaged total value y of the second indicator; if the analysis object is The covariance between the bucket-averaged total index values of the two to-be-analyzed indicators in the analysis object is Cov(z,y), that is, the covariance between the bucket-averaged total index value y of the second indicator and the bucket-averaged total index value z of the number of data records.

[0105] In one possible implementation, Figure 7 As shown, step S2113 may include:

[0106] S21131, for each indicator to be analyzed, calculate the statistical results of the indicator to be analyzed in each data bucket; based on the statistical results of the indicator to be analyzed in each data bucket and the total number of data buckets, calculate the bucket average indicator total value of the indicator to be analyzed.

[0107] The server can directly calculate the statistics of the metric to be analyzed within the sample set of data records. Then, based on these statistics and the total data in the data bucket, it calculates the bucket-averaged metric value for the metric to be analyzed. However, in scenarios with large data volumes, the sample set of data records can be extremely large, requiring the server to have high computing power. Therefore, directly calculating statistics based on the sample set of data records is more suitable for scenarios with smaller data volumes. For scenarios with large data volumes, the server can determine the metric value based on the corresponding metric value in each data bucket.

[0108] When calculating the statistical results of the indicator to be analyzed in each data bucket, the server may include: for each data bucket, calculating the sum of the indicator values of the indicator to be analyzed in the data bucket to obtain the statistical results of the indicator to be analyzed in the data bucket. Accordingly, calculating the bucket-averaged total indicator value of the indicator to be analyzed based on the statistical results of the indicator to be analyzed in each data bucket and the total number of data buckets may include: determining the bucket-averaged total indicator value of the indicator to be analyzed as the ratio of the sum of the statistical results of the indicator to be analyzed in each data bucket to the total number of data buckets.

[0109] In one example, Figure 3 As an example, if we use BH j Indicates the statistical result of the first indicator in the j-th data bucket, BW j Indicates the statistical results of the second indicator in the j-th data bucket, BN j represents the statistical result of the number of data records in the j-th data bucket, then BH j and BW j Can be expressed as and Correspondingly, the bucket-averaged total index value x of the first indicator, the bucket-averaged total index value y of the second indicator, and the bucket-averaged total index value z of the number of data records can be expressed as follows: as well as

[0110] S21133: Obtain the covariance between the bucket-averaged total values of the two indicators to be analyzed based on the bucket-averaged total values of the indicators to be analyzed and the statistical results of the indicators to be analyzed in the data bucket.

[0111] When calculating the covariance between the bucket-averaged total values of two metrics to be analyzed, the server first calculates the difference between the statistical result of each metric to be analyzed and the bucket-averaged total value of the metric to be analyzed for each data bucket as the difference-mean value for that metric. The product of the difference-mean values of each metric to be analyzed is then used as the target product for that data bucket. The covariance between the bucket-averaged total values of the metric to be analyzed is then calculated based on the sum of the target products for each data bucket and the total number of data buckets.

[0112] In one example, Figure 3 Taking the data bucket shown as an example, assuming that the covariance between the bucket-averaged index total value x of the first indicator and the bucket-averaged index total value y of the second indicator in the analysis object is calculated, that is, Cov(x,y), for the j-th data bucket, the difference mean of the first indicator can be expressed as BH j -x, the difference mean of the second indicator can be expressed as BW j -y, and then we can get the target product of the j-th data bucket, which can be expressed as (BH j -x)(BW j -y). Then, the covariance between the bucket-averaged total value of the first indicator and the bucket-averaged total value of the second indicator can be calculated by the following formula:

[0113]

[0114]

[0115] Among them, x represents the total value of the first indicator in the bucket, y represents the total value of the second indicator in the bucket, BH j Indicates the statistical result of the first indicator in the j-th data bucket, BW j represents the statistical result of the second indicator in the j-th data bucket, and B represents the total number of data buckets.

[0116] S2115: Obtain the covariance between the first analysis target and the second analysis target based on the calculation result.

[0117] In one example, as described in step S21113 above, the covariance between the first analysis target and the second analysis target can be expressed as:

[0118]

[0119] After obtaining the values of Cov(x,y), Cov(z,z), Cov(z,y), and Cov(x,z), the value of Cov(f1,f2) can be determined.

[0120] S213: Determine an analysis result of the data record according to the covariance between the first analysis target and the second analysis target.

[0121] The covariance between the first analysis target and the second analysis target indicates the degree of joint change of the first analysis target and the second analysis target or the overall error between the first analysis target and the second analysis target. The covariance between the first analysis target and the second analysis target can be used as the analysis result of the data record, or the analysis result of the data record can be determined based on the covariance between the first analysis target and the second analysis target.

[0122] In one example, in an AB testing scenario, the product characteristics of Group A and Group B differ to some extent. The first analysis objective and the second analysis objective correspond to the product performance of Group A and Group B, respectively. Based on the covariance between the first analysis objective and the second analysis objective, it is possible to determine which group's product characteristics are superior, thereby obtaining the analysis results for the AB testing scenario. For example, if the covariance between the first analysis objective and the second analysis objective is within a preset range, it can be determined that Group A's product characteristics are superior.

[0123] In an embodiment of the present application, the covariance between the first analysis target and the second analysis target can be calculated based on the total value of the bucket average indicator, and the total value of the bucket average indicator is only related to the total number of data buckets and the data in the data buckets, and the data in each data bucket is only a part of the entire data record set, so that when the server calculates the covariance between the first analysis target and the second analysis target, it achieves the effect of dimensionality reduction, thereby being able to quickly obtain the analysis results of the data records.

[0124] As described in the aforementioned step S203, when the server accesses the data record storage address and obtains the data record set, it can determine all the data records in the data record storage address as the data record set, or it can select a preset number of data records from all the data records as the data record set. In this scenario, the data records stored in the data record storage address are data that the server can directly bucket, that is, each data record contains a first indicator and a second indicator. In other scenarios, the data stored in the data record storage address needs to be merged, filtered, or other column processing before it can be bucketed; or, there can be at least two data record storage addresses, each of which corresponds to the indicator data of an indicator.

[0125] In view of this, in one possible implementation, Figure 8 As shown, step S201 may include:

[0126] S2011, obtaining a data record storage address corresponding to test data, where the test data includes first test data and second test data.

[0127] The first test data represents test data for a first indicator, and the second test data represents test data for a second indicator. It is understood that the data record storage address corresponding to the first test data and the data record storage address corresponding to the second test data can be the same or different. For example, in an AB testing scenario, the first test data can be experimental group data, and the second test data can be control group data. The experimental group data and the control group data can be stored in the same or different memory or database tables.

[0128] Accordingly, step S203 may include:

[0129] S2031, accessing a data record storage address corresponding to the first test data to obtain at least one first data record corresponding to the first test data; and accessing a data record storage address corresponding to the second test data to obtain at least one second data record corresponding to the second test data;

[0130] S2033: Merge the at least one first data record and the at least one second data record to obtain a data record set.

[0131] In the embodiment of the present application, the merging process is mainly used to configure the first data record and the second data record into a pair to obtain a data record in the data record set.

[0132] In one possible implementation, Figure 9 As shown, step S2033 may include:

[0133] S20331, obtain the target record quantity;

[0134] The target number of records represents the number of data records to be analyzed. The total number of first data records retrieved by the server may be inconsistent with the total number of second data records. To ensure that each data record contains the first and second indicators, the server may determine the minimum value between the total number of first data records and the total number of second data records as the target number of records. In specific implementations, a preset number of records may also be used as the target number of records.

[0135] S20333: Randomly select a target number of first data records from the at least one first data record to obtain a first data record set; and randomly select a target number of second data records from the at least one second data record to obtain a second data record set.

[0136] S20335: For each first data record in the first data record set, combine the first data record with a second data record arbitrarily selected from the second data record set to form a data record in the data record set.

[0137] For example, assuming the target number of records is N, if {H1,H2,…,H N} represents the first data record set, and {W1,W2,…,W N} represents the second data record set, then for any H i (1≤i≤N), there exists a unique W j (1≤j≤N) such that {H i ,W j} constitutes a data record in the data record set.

[0138] The following uses test data as an example to illustrate the data record analysis method of the embodiment of the present application. The test data includes the height and weight of N users, that is, the first indicator is height, the second indicator is weight, and each data record corresponds to the test data of one user. i Indicates the height of the i-th user, using W i represents the weight of the i-th user. Assume that the first analysis target is the average weight and Indicates that the second analysis target is the average height and is expressed as In the prior art, if we want to analyze the correlation between average weight and average height, we usually calculate the covariance between average weight and average height, that is:

[0139]

[0140] Among them, Cov(H,W) represents the covariance between height and weight, which can be obtained by Calculated.

[0141] It can be seen that when the number of users N is very large, the corresponding storage and computing overhead is also very large, which makes it difficult to estimate the covariance between average height and average weight in large data scenarios.

[0142] The data record analysis method of the embodiment of the present application first randomly distributes the test data of N users into B data buckets based on the bucketing method, and then uses the data in each data bucket for calculation, that is, converts the original data into bucketed data. The data in all data buckets constitute the original data, but during the calculation, only the data allocated in each data bucket needs to be considered, so that the amount of data in each data bucket after bucketing is greatly reduced compared to the entire data record set, thereby achieving the effect of dimensionality reduction of the number of data records.

[0143] As described in step S2113 above, using the data record analysis method of the embodiment of the present application, the covariance between the calculated average height and average weight can be expressed as:

[0144]

[0145] Here, z represents the bucket-averaged total value of the number of users, x represents the bucket-averaged total value of the height, and y represents the bucket-averaged total value of the weight. As can be seen from the above formula, the covariance between average height and average weight is only related to the bucket-averaged total value. Calculating the covariance between bucket-averaged total values significantly reduces the computational complexity compared to calculating the covariance between the entire raw data. This allows covariance estimation to quickly generate analytical results for data records in large data volumes.

[0146] It can be seen from the technical solutions provided by the above embodiments that the data record analysis method provided by the embodiments of the present application uses a bucketing method to randomly assign each data record in the data record set to a data bucket, thereby achieving the effect of reducing the dimensionality of the original data record set, so that when calculating the covariance between the first analysis target and the second analysis target, the bucket average indicator total value can be directly used for analysis, thereby improving the calculation and storage rate of the data record analysis, so that the analysis results of the data records can be obtained quickly, and solving the problem of using covariance estimation to quickly obtain the analysis results of the data records in large data volume scenarios.

[0147] Based on the same inventive concept as the above method embodiment, the embodiment of the present application also provides a data recording and analysis device, such as Figure 10 As shown, the apparatus 1000 may include:

[0148] The storage address acquisition module 1010 is configured to acquire the data record storage address in response to the data record analysis instruction;

[0149] A data record determination module 1020 is configured to access a data record storage address to obtain a data record set, wherein each data record in the data record set includes a first indicator and a second indicator;

[0150] Bucket processing module 1030, for randomly assigning each data record in the data record set to a data bucket;

[0151] An analysis indicator determination module 1040 is configured to determine the first indicator, the second indicator, and the number of data records as indicators to be analyzed;

[0152] Analysis target acquisition module 1050 is configured to acquire a first analysis target for analyzing a first data relationship between bucket-averaged total values of at least two indicators to be analyzed; and acquire a second analysis target for analyzing a second data relationship between bucket-averaged total values of at least two indicators to be analyzed;

[0153] A covariance determination module 1060 is configured to determine the covariance between the first analysis target and the second analysis target based on the bucket average index total value;

[0154] The analysis result determination module 1070 is configured to determine an analysis result of the data record based on the covariance between the first analysis target and the second analysis target.

[0155] In one possible implementation, the covariance determination module 1060 may include:

[0156] a first analysis object determination unit, configured to determine an analysis object according to a first analysis target and a second analysis target;

[0157] A first covariance calculation unit is used to calculate the covariance between the bucket average index total values of two indicators to be analyzed in the analysis object;

[0158] The second covariance calculation unit is configured to obtain the covariance between the first analysis target and the second analysis target according to the calculation result.

[0159] In one possible implementation, the first covariance calculation unit may include:

[0160] The analysis indicator statistics unit is used to calculate the statistical results of the indicator to be analyzed in each data bucket for each indicator to be analyzed; and calculate the bucket average indicator value of the indicator to be analyzed based on the statistical results of the indicator to be analyzed in each data bucket and the total number of data buckets;

[0161] The covariance factor calculation unit is used to obtain the covariance between the bucket average indicator total values of two indicators to be analyzed based on the bucket average indicator total values of each indicator to be analyzed and the statistical results of each indicator to be analyzed in the data bucket.

[0162] In one possible implementation, the first analysis object determination unit may include:

[0163] a gradient operation unit, configured to perform a gradient operation and a difference operation on the first analysis target to obtain a third analysis target; and perform a gradient operation and a difference operation on the second analysis target to obtain a fourth analysis target;

[0164] The second analysis object determination unit is configured to determine an analysis object according to the third analysis object and the fourth analysis object.

[0165] In one possible implementation, the gradient operation unit may include a first gradient operation unit. Specifically, the first gradient operation unit may include:

[0166] a first indicator vector determining unit, configured to obtain each to-be-analyzed indicator corresponding to the first data relationship, and determine a first indicator vector according to the bucket-averaged indicator total value of each to-be-analyzed indicator corresponding to the first data relationship;

[0167] a first mean vector determining unit, configured to determine the expectation of the first indicator vector as a first mean vector;

[0168] a first gradient vector calculation unit, configured to obtain a first gradient vector by calculating a partial derivative of the first analysis target with respect to the first index vector;

[0169] A first difference vector calculation unit, configured to calculate the difference between the first index vector and the first mean vector to obtain a first difference vector;

[0170] a first expected vector calculation unit, configured to determine an expectation of the first gradient vector as a first expected vector;

[0171] The target determination unit is used to determine the product of the transpose of the first expected vector and the first difference vector as a third analysis target.

[0172] In one possible implementation, the storage address acquisition module 1010 may include:

[0173] The address acquisition unit is used to obtain the data record storage address corresponding to the test data, the test data includes first test data and second test data, the first test data represents the test data of the first indicator, and the second test data represents the test data of the second indicator.

[0174] Accordingly, the data record determination module 1020 may include:

[0175] a first data record determining unit, configured to access a data record storage address corresponding to the first test data to obtain at least one first data record corresponding to the first test data;

[0176] a second data record determining unit, configured to access a data record storage address corresponding to the second test data to obtain at least one second data record corresponding to the second test data;

[0177] The data record determining unit is configured to merge at least one first data record and at least one second data record to obtain a data record set.

[0178] In one possible implementation, the data record determination unit may include:

[0179] A quantity determination unit, used to obtain the target record quantity;

[0180] a first data screening unit, configured to randomly select a target number of first data records from at least one first data record to obtain a first data record set;

[0181] a second data screening unit, configured to randomly select a target number of second data records from at least one second data record to obtain a second data record set;

[0182] The data pairing unit is configured to combine each first data record in the first data record set with a second data record arbitrarily selected from the second data record set to form a data record in the data record set.

[0183] It should be noted that the apparatus provided in the above embodiments, when implementing its functions, is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0184] An embodiment of the present application also provides an electronic device, which includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded by the processor and executes the data recording and analysis method provided by the above method embodiment.

[0185] Further, Figure 11 A schematic diagram of the hardware structure of a device for implementing the method provided in the embodiment of the present application is shown, and the device may participate in or include the apparatus or system provided in the embodiment of the present application. Figure 11 As shown, the device 11 may include one or more (illustrated as 1102a, 1102b, ..., 1102n in the figure) processors 1102 (the processor 1102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 1104 for storing data, and a transmission device 1106 for communication functions. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of the I / O interface), a network interface, a power supply and / or a camera. It will be understood by those skilled in the art that Figure 11 The structure shown is only for illustration and does not limit the structure of the above electronic device. Figure 11 More or fewer components than shown, or with Figure 11 Different configurations shown.

[0186] It should be noted that the one or more processors 1102 and / or other data processing circuits described above may generally be referred to herein as "data processing circuitry". The data processing circuitry may be embodied in whole or in part as software, hardware, firmware, or any other combination thereof. In addition, the data processing circuitry may be a single independent processing module, or may be incorporated in whole or in part into any of the other components of the device 11 (or mobile device). As described in the embodiments of the present application, the data processing circuitry serves as a processor control (e.g., selection of a variable resistor terminal path connected to an interface).

[0187] The memory 1104 can be used to store software programs and modules of application software, such as program instructions / data storage devices corresponding to the methods described in the embodiments of the present application. The processor 1102 executes various functional applications and data processing by running the software programs and modules stored in the memory 1104, that is, implementing the above-mentioned data recording and analysis method. The memory 1104 may include a high-speed random access memory and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 1104 may further include a memory remotely located relative to the processor 1102, and these remote memories may be connected to the device 11 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0188] Transmission device 1106 is configured to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by the communications provider of device 11. In one embodiment, transmission device 1106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, transmission device 1106 may be a radio frequency (RF) module, which is configured to communicate with the Internet wirelessly.

[0189] The display may be, for example, a touch screen liquid crystal display (LCD) that enables a user to interact with a user interface of the device 11 (or mobile device).

[0190] An embodiment of the present application also provides a computer-readable storage medium, which stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by a processor to implement the data recording and analysis method provided by the above method embodiment.

[0191] Optionally, in this embodiment, the computer-readable storage medium may be located in at least one of a plurality of network servers in a computer network. Optionally, in this embodiment, the storage medium may include, but is not limited to, various media capable of storing program code, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk, or an optical disk.

[0192] The present application also provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the data record analysis method provided in the method embodiment described above.

[0193] It should be noted that the order of the embodiments of the present application described above is for descriptive purposes only and does not represent the superiority or inferiority of the embodiments. The above description is of specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0194] The various embodiments in this specification are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the device and electronic device embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0195] The above description has fully disclosed the specific embodiments of this application. It should be noted that any changes made by those skilled in the art to the specific embodiments of this application do not depart from the scope of the claims of this application. Accordingly, the scope of the claims of this application is not limited to the above specific embodiments.

Claims

1. A data record analysis method, characterized in that: The method comprises: In response to the data record analysis instruction, obtaining a data record storage address; Accessing the data record storage address to obtain a data record set, wherein each data record in the data record set includes a first indicator and a second indicator; Randomly assigning each data record in the data record set to a data bucket; Determining the first indicator, the second indicator, and the number of data records as indicators to be analyzed; Obtaining a first analysis target, the first analysis target being used to analyze a first data relationship between a bucket-averaged total index value of the first indicator and a bucket-averaged total index value of the number of data records; and obtaining a second analysis target, the second analysis target being used to analyze a second data relationship between a bucket-averaged total index value of the second indicator and a bucket-averaged total index value of the number of data records; determining a covariance between the first analysis objective and the second analysis objective; An analysis result of the data record is determined based on the covariance between the first analysis target and the second analysis target.

2. The method according to claim 1, characterized in that The determining the covariance between the first analysis target and the second analysis target includes: Calculate the covariance between the bucket-averaged index total value of the first indicator and the bucket-averaged index total value of the number of data records, the covariance between the bucket-averaged index total value of the second indicator and the bucket-averaged index total value of the number of data records, the covariance between the bucket-averaged index total value of the first indicator and the bucket-averaged index total value of the second indicator, and the covariance of the bucket-averaged index total value of the number of data records itself; According to the calculation result, the covariance between the first analysis target and the second analysis target is obtained.

3. The method according to claim 2, characterized in that The method further comprises: For each of the indicators to be analyzed, calculate the statistical results of the indicator to be analyzed in each of the data buckets; calculate the bucket-average indicator value of the indicator to be analyzed based on the statistical results of the indicator to be analyzed in each of the data buckets and the total number of the data buckets; The covariance between the bucket-averaged total index values of two indicators to be analyzed is obtained according to the bucket-averaged total index values of each indicator to be analyzed and the statistical results of each indicator to be analyzed in the data bucket.

4. The method according to claim 2 or 3, characterized in that The method further comprises: performing a gradient operation and a difference operation on the first analysis target to obtain a third analysis target; and performing the gradient operation and the difference operation on the second analysis target to obtain a fourth analysis target; A corresponding analysis object is determined according to the third analysis target and the fourth analysis target.

5. The method according to claim 4, characterized in that The performing gradient operation and difference operation on the first analysis target to obtain a third analysis target includes: Obtaining each of the to-be-analyzed indicators corresponding to the first data relationship, and determining a first indicator vector according to the bucket-averaged indicator total value of each of the to-be-analyzed indicators corresponding to the first data relationship; Determine the expectation of the first indicator vector as a first mean vector; Taking a partial derivative of the first analysis target with respect to the first indicator vector to obtain a first gradient vector; Calculating the difference between the first indicator vector and the first mean vector to obtain a first difference vector; determining an expectation of the first gradient vector as a first expectation vector; The product of the transpose of the first expected vector and the first difference vector is determined as the third analysis target.

6. The method according to claim 1, characterized in that The obtaining of the data record storage address includes: Obtaining a data record storage address corresponding to test data, the test data including first test data and second test data, the first test data representing test data of the first indicator, and the second test data representing test data of the second indicator; Accordingly, accessing the data record storage address to obtain a data record set includes: Accessing a data record storage address corresponding to the first test data to obtain at least one first data record corresponding to the first test data; and accessing a data record storage address corresponding to the second test data to obtain at least one second data record corresponding to the second test data; The at least one first data record and the at least one second data record are merged to obtain the data record set.

7. The method according to claim 6, characterized in that The merging of the at least one first data record and the at least one second data record to obtain the data record set includes: Get the target record quantity; Randomly selecting the target number of first data records from the at least one first data record to obtain a first data record set; and randomly selecting the target number of second data records from the at least one second data record to obtain a second data record set; For each first data record in the first data record set, the first data record and one second data record arbitrarily selected from the second data record set are combined to form a data record in the data record set.

8. A data recording and analysis device, characterized in that: The device comprises: a storage address acquisition module, configured to acquire a data record storage address in response to a data record analysis instruction; a data record determination module, configured to access the data record storage address to obtain a data record set, wherein each data record in the data record set includes a first indicator and a second indicator; A bucket processing module, configured to randomly assign each data record in the data record set to a data bucket; an analysis indicator determination module, configured to determine the first indicator, the second indicator, and the number of data records as indicators to be analyzed; an analysis target acquisition module, configured to acquire a first analysis target for analyzing a first data relationship between a bucket-averaged total index value of the first indicator and a bucket-averaged total index value of the number of data records; and to acquire a second analysis target for analyzing a second data relationship between a bucket-averaged total index value of the second indicator and a bucket-averaged total index value of the number of data records; a covariance determination module, configured to determine the covariance between the first analysis target and the second analysis target; An analysis result determination module is used to determine an analysis result of the data record according to the covariance between the first analysis target and the second analysis target.

9. The device according to claim 8, characterized in that The covariance determination module includes: a first covariance calculation unit, configured to respectively calculate the covariance between the bucket-averaged index total value of the first indicator and the bucket-averaged index total value of the number of data records, the covariance between the bucket-averaged index total value of the second indicator and the bucket-averaged index total value of the number of data records, the covariance between the bucket-averaged index total value of the first indicator and the bucket-averaged index total value of the second indicator, and the covariance of the bucket-averaged index total value of the number of data records itself; The second covariance calculation unit is configured to obtain the covariance between the first analysis target and the second analysis target according to a calculation result.

10. The device according to claim 9, characterized in that The device is also used for: For each of the indicators to be analyzed, calculate the statistical results of the indicator to be analyzed in each of the data buckets; calculate the bucket-average indicator value of the indicator to be analyzed based on the statistical results of the indicator to be analyzed in each of the data buckets and the total number of the data buckets; The covariance between the bucket-averaged total index values of two indicators to be analyzed is obtained according to the bucket-averaged total index values of each indicator to be analyzed and the statistical results of each indicator to be analyzed in the data bucket.

11. The device according to claim 8 or 9, characterized in that The device is also used to perform gradient operation and difference operation on the first analysis target to obtain a third analysis target; and perform the gradient operation and difference operation on the second analysis target to obtain a fourth analysis target; and determine the corresponding analysis object according to the third analysis target and the fourth analysis target.

12. The device according to claim 11, characterized in that The performing gradient operation and difference operation on the first analysis target to obtain a third analysis target includes: Obtaining each of the to-be-analyzed indicators corresponding to the first data relationship, and determining a first indicator vector according to the bucket-averaged indicator total value of each of the to-be-analyzed indicators corresponding to the first data relationship; Determine the expectation of the first indicator vector as a first mean vector; Taking a partial derivative of the first analysis target with respect to the first indicator vector to obtain a first gradient vector; Calculating the difference between the first indicator vector and the first mean vector to obtain a first difference vector; determining an expectation of the first gradient vector as a first expectation vector; The product of the transpose of the first expected vector and the first difference vector is determined as the third analysis target.

13. The device according to claim 8, characterized in that The storage address acquisition module includes: An address acquisition unit, configured to acquire a data record storage address corresponding to test data, the test data comprising first test data and second test data, the first test data representing test data of the first indicator, and the second test data representing test data of the second indicator; Accordingly, the data record determination module includes: a first data record determining unit, configured to access a data record storage address corresponding to the first test data to obtain at least one first data record corresponding to the first test data; a second data record determining unit, configured to access a data record storage address corresponding to the second test data to obtain at least one second data record corresponding to the second test data; The data record determining unit is configured to merge the at least one first data record and the at least one second data record to obtain the data record set.

14. The device according to claim 13, characterized in that The data record determination unit includes: A quantity determination unit, used to obtain the target record quantity; a first data screening unit, configured to randomly select the target number of first data records from the at least one first data record to obtain a first data record set; a second data screening unit, configured to randomly select the target number of second data records from the at least one second data record to obtain a second data record set; The data pairing unit is configured to, for each first data record in the first data record set, combine the first data record with a second data record arbitrarily selected from the second data record set to form a data record in the data record set.

15. An electronic device, characterized in that: The device includes a processor and a memory, wherein the memory stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded by the processor and executes the data record analysis method according to any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that The computer-readable storage medium stores at least one instruction or at least one program, and the at least one instruction or the at least one program is loaded and executed by a processor to implement the data record analysis method according to any one of claims 1 to 7.

Images