A data intelligence sharing method for building a smart sharing system

Abstract

This invention relates to the field of data classification technology, specifically to a data intelligent sharing method for building a smart sharing system. This method splits the total historical business data into different subsets using each conditional attribute. Based on the differences in the disorder of indicator characteristics of the indicator attributes in the total historical business data and the subsets of each conditional attribute, as well as the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, an initial priority index is obtained for each conditional attribute. The initial priority index is then adjusted by combining the influence of each conditional attribute on the attribute of interest and the correlation between each conditional attribute and other conditional attributes, thereby constructing a decision tree to classify and share the total historical business data. This invention combines the information gain, importance, and correlation of different conditional attributes brought by conditional attributes when splitting the dataset to construct the decision tree, improving the effectiveness of the decision tree in classifying based on the attribute of interest.

Description

Technical Field

[0001] This invention relates to the field of data classification technology, and more specifically to a data intelligent sharing method for the construction of a smart sharing system. Background Technology

[0002] The global wave of informatization has invigorated the construction of data sharing systems, and the development of various shared databases has improved the scientific rigor, accuracy, and efficiency of data analysis. Data sharing, as a crucial means of driving economic and social development, is of great significance for promoting innovation, improving efficiency, and achieving sustainable development.

[0003] Some companies store directly collected data in databases or data platforms to share relevant company data. However, this directly collected data is not fully explored and utilized, thus failing to effectively support business analysis and decision-making. To fully extract valuable information from company-related data, attributes are typically extracted, and decision trees are constructed to classify the data based on these attributes. Existing methods determine the priority of each attribute in decision tree construction based on the information gain it brings when segmenting the dataset. However, complex relationships often exist between different attributes in company-related data, and the degree of correlation between different attributes and important information varies. For example, insurance companies focus on profits, and expense ratios and loss ratios jointly affect profitability. Even with high premium income, a simultaneous increase in both expense ratio and loss ratio can lead to a decrease in net profit. Furthermore, the varying degrees of correlation between insurance income, loss ratio, and profit result in poor performance of decision trees based on profit information, leading to low accuracy in interpreting the profitability of insurance companies. Summary of the Invention

[0004] To address the technical problem that determining the priority of a decision tree based solely on the information gain of each attribute when segmenting a dataset leads to poor data classification performance, this invention aims to provide a data intelligent sharing method for building a smart sharing system. The specific technical solution adopted is as follows:

[0005] This invention proposes a data intelligent sharing method for the construction of a smart sharing system, the method comprising:

[0006] Obtain total historical business data, which includes historical business data from several years. Extract business features from the total historical business data and determine the indicator attributes, attention attributes, and several conditional attributes among the business features.

[0007] The total historical business data is split using each conditional attribute to obtain different subsets for each conditional attribute. Based on the differences in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subsets of each conditional attribute, as well as the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, the initial priority indicators for each conditional attribute are obtained.

[0008] Based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes, the initial priority indicators are adjusted, and a decision tree is constructed using the adjusted initial priority indicators to classify and share the total historical business data.

[0009] Furthermore, obtaining the initial priority index for each conditional attribute includes:

[0010] Based on the difference in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subset of each conditional attribute, obtain the information gain of each conditional attribute.

[0011] Based on the similarity of the changing trends of historical business data for each conditional attribute and the attribute of interest, obtain the collaborative indicators for each conditional attribute and the attribute of interest.

[0012] Based on the information gain and the synergy index, an initial priority index is obtained for each conditional attribute; both the information gain and the synergy index are positively correlated with the initial priority index.

[0013] Furthermore, obtaining the information gain of each conditional attribute includes:

[0014] Obtain the information entropy of the indicator characteristics of the indicator attributes in the total historical business data, and denote it as the total information entropy;

[0015] Obtain the information entropy of the indicator features of the indicator attributes in the historical business data of each subset of each conditional attribute, and denote it as the sub-information entropy of the corresponding subset; take the proportion of the annual quantity in the historical business data of each subset of each conditional attribute to the annual quantity in the total historical business data as the quantity indicator of each subset of each conditional attribute.

[0016] Based on the sub-information entropy and the quantity index of all subsets of each conditional attribute, obtain the overall uncertainty index of each conditional attribute;

[0017] The difference between the total information entropy and the overall uncertainty index is normalized to obtain the information gain of each conditional attribute.

[0018] Furthermore, obtaining the collaborative metrics for each conditional attribute and the attribute of interest includes:

[0019] The conditional attributes and the attributes of interest are recorded as analysis attributes. The historical business data of each analysis attribute are arranged sequentially to obtain the business data sequence of each analysis attribute.

[0020] Obtain the second-order difference sequence of the business data sequence for each analysis attribute; calculate the mean of the absolute values ​​of the differences between the elements with the same index in the second-order difference sequence of each conditional attribute and the attribute of interest, and use it as an indicator of the change difference between each conditional attribute and the attribute of interest.

[0021] Obtain the correlation coefficient of the business data sequence for each conditional attribute and the attribute of interest. Based on the correlation coefficient and the change difference index, obtain the coordination index for each conditional attribute and the attribute of interest. The correlation coefficient and the coordination index are positively correlated, and the change difference index and the coordination index are negatively correlated.

[0022] Furthermore, the initial priority index is adjusted based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes. A decision tree is then constructed using the adjusted initial priority index to classify and share the total historical business data, including:

[0023] The initial priority index is adjusted based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes to obtain the final priority index for each conditional attribute.

[0024] Arrange all conditional attributes in descending order of their final priority indices to obtain an initial priority sequence; select the first few conditional attributes from the initial priority sequence to form the final priority sequence.

[0025] According to the order of the condition attributes in the final priority sequence, a decision tree is constructed using the Hunt algorithm to obtain the classification results of the total historical business data, and the obtained classification results are stored in the database.

[0026] Furthermore, obtaining the final priority index for each conditional attribute includes:

[0027] Obtain the impact metrics of each conditional attribute on the attribute of interest;

[0028] Obtain the correlation coefficient of each conditional attribute with the business data sequence of other conditional attributes, and take the mean of the correlation coefficients of each conditional attribute with the business data sequence of other conditional attributes as the overall correlation value of each conditional attribute with other conditional attributes.

[0029] The overall correlation value is negatively correlated, and the product of the mapping result and the influence index of each conditional attribute is normalized to obtain the priority construction coefficient of each conditional attribute.

[0030] The initial priority index is weighted using the priority construction coefficient to obtain the final priority index for each conditional attribute.

[0031] Furthermore, the step of obtaining the impact index of each conditional attribute on the attribute of interest includes:

[0032] Linear analysis is performed on all conditional and interest attributes to construct an initial interest function, which contains several regression coefficients.

[0033] Substitute the historical business data of all conditional attributes and the historical business data of the attention attributes for each year into the initial attention function to obtain the attention fit value for each year.

[0034] Based on the differences between the historical business data of the attention attributes for all years and the fitted values ​​of the attention, an error function is obtained; the value of the regression coefficient corresponding to the minimum value of the error function is taken as the optimal value of each regression coefficient in the initial attention function, and the optimal value of the regression coefficient corresponding to each conditional attribute is taken as the influence index of each conditional attribute.

[0035] Furthermore, the step of splitting the total historical business data using each conditional attribute to obtain different subsets for each conditional attribute includes:

[0036] The product of the maximum value in the historical business data of each conditional attribute and the preset weight threshold is used as the standard segmentation threshold for each conditional attribute.

[0037] Based on the standard segmentation threshold, the historical business data of each conditional attribute is divided into two categories; the set consisting of the historical business data of each year to which each data belongs in each category of historical business data for each conditional attribute is used as a subset of each conditional attribute.

[0038] Further, obtaining the overall uncertainty index for each conditional attribute based on the sub-information entropy and the quantity index of all subsets of each conditional attribute includes:

[0039] The product of the sub-information entropy and the quantity index for each subset of each conditional attribute is used as the local uncertainty index for the corresponding subset; the sum of the local uncertainty indices for all subsets of each conditional attribute is used as the overall uncertainty index for each conditional attribute.

[0040] Furthermore, the correlation coefficient is the Pearson correlation coefficient.

[0041] The present invention has the following beneficial effects:

[0042] In this embodiment of the invention, the difference in the degree of disorder of the indicator features of the indicator attributes in the total historical business data and the subset of each conditional attribute presents the information gain after dividing the total historical business data using each conditional attribute. The similarity of the changing trends of the historical business data of each conditional attribute and the attention attribute is used to analyze the influence of each conditional attribute on the key attention information. The initial priority index is obtained by combining the analysis of the two factors, which improves the accuracy of the priority information for decision analysis of each conditional attribute. Since the degree of correlation between different attributes in the business data will affect the priority of the conditional attributes in the process of building the decision tree, in order to build a better decision tree, it is necessary to consider the degree of correlation between each conditional attribute and other conditional attributes, and combine the degree of influence of each conditional attribute on the attention attribute to adjust the initial priority index. The decision tree is built using the adjusted initial priority index. Considering the attention attribute, i.e., the key attention information, the classification effect of the decision tree on the total historical business data is improved, the accuracy of the interpretation of the key attention information is improved, and the problem of poor classification effect of decision trees built only based on information gain based on key attention information is effectively solved. Attached Figure Description

[0043] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A flowchart illustrating the steps of a data intelligent sharing method for building a smart sharing system, provided in one embodiment of the present invention;

[0045] Figure 2 A flowchart illustrating the steps of an initial priority index acquisition method provided in one embodiment of the present invention;

[0046] Figure 3 This is a system structure diagram of a data intelligent sharing system for building a smart sharing system, provided in one embodiment of the present invention;

[0047] Figure 4 This is a schematic diagram of a computer device for building a smart sharing system, provided as an embodiment of the present invention. Detailed Implementation

[0048] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a data intelligent sharing method for building a smart sharing system based on the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0049] 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 invention pertains.

[0050] The following description, in conjunction with the accompanying drawings, details a specific scheme for a data intelligent sharing method for building a smart sharing system provided by the present invention.

[0051] Example 1:

[0052] This invention proposes a data intelligent sharing method for the construction of a smart sharing system. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of a data intelligent sharing method for building a smart sharing system, provided by an embodiment of the present invention. The method includes:

[0053] Step S1: Obtain total historical business data, which includes historical business data from several years. Extract business features from the total historical business data and determine the indicator attributes, attention attributes, and several conditional attributes in the business features.

[0054] This solution uses the financial data of a property insurance company as an example for analysis. The total historical business data is the financial data of a property insurance company with different attributes over several years. The total historical business data can be obtained from the corresponding business database, which is a pre-established database specifically for storing business data.

[0055] Business characteristics are extracted from business data, including insurance revenue, loss ratio, expense ratio, net profit, auto insurance loss ratio, and profitability. The loss ratio refers to the proportion of claims paid out by a property insurance company to its premium income, while the expense ratio refers to the proportion of various expenses incurred by a property insurance company for operating and selling insurance to its premium income.

[0056] To uncover high-value data from financial data, this solution requires constructing a suitable decision tree to analyze which attributes primarily influence the profitability of a property insurance company, thereby gaining an overall understanding of the company's profit situation.

[0057] The indicator attribute is a decision attribute, which is determined based on the condition attributes. This solution analyzes which attributes mainly affect the profitability of a property insurance company, and since the company's profitability can be measured by net profit, the indicator attribute in this embodiment is whether it is profitable, the focus attribute is net profit, and the condition attributes include: insurance revenue, loss ratio, expense ratio, and auto insurance loss ratio.

[0058] Different conditional attributes have different impacts on company profitability. Conditional attributes need to be used to make decisions on total historical business data in an appropriate order to maximize the purity of decision categories, thereby more accurately interpreting the company's profitability and obtaining high-value financial data.

[0059] As an example, the business data table consisting of total historical business data is shown in Table 1:

[0060] Table 1, Business Data Table

[0061]

[0062]

[0063] It should be noted that all the data in Table 1 are collectively referred to as the total historical business data; the historical business data for each attribute is all the data in the column containing that attribute in Table 1. As an example, the historical business data for insurance revenue includes: 3800, 6500, 4500, 1800, 2500, 3200, 1200, and 1600; the historical business data for each year is the data in the same row of Table 1.

[0064] The data of different attributes in the total historical business data have different dimensions and orders of magnitude. Directly using historical business data for calculations may highlight some indicators with larger orders of magnitude while excluding some indicators with smaller orders of magnitude. Therefore, it is necessary to normalize each business data in the total historical business data to obtain normalized business data. In this embodiment of the invention, min-max normalization is used to normalize each business data. Normalization methods such as Norm functions and function transformations can also be used, and no limitation is made here. Subsequent steps will all use normalized business data for analysis.

[0065] Step S2: Split the total historical business data using each conditional attribute to obtain different subsets for each conditional attribute; based on the differences in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subsets of each conditional attribute, as well as the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, obtain the initial priority indicator for each conditional attribute.

[0066] To simplify the analysis of information gain for different conditional attributes, this embodiment adopts a binary split for each conditional attribute in decision splitting, that is, after splitting the total historical business data based on each conditional attribute, two subsets are obtained.

[0067] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining a subset of each conditional attribute includes: multiplying the maximum value in the historical business data of each conditional attribute by a preset weight threshold as a standard segmentation threshold for each conditional attribute; dividing the historical business data of each conditional attribute into two categories based on the standard segmentation threshold; and using the set of historical business data of each year to which each data belongs in each category of historical business data of each conditional attribute as a subset of each conditional attribute.

[0068] It should be noted that for each conditional attribute's historical business data, historical business data exceeding the standard segmentation threshold is classified as the first type of historical business data for each conditional attribute, while historical business data less than or equal to the standard segmentation threshold is classified as the second type of historical business data. In this embodiment of the invention, the preset weight threshold is taken as an empirical value of 0.3, which implementers can set according to specific circumstances. In Table 1, the data in each row of historical business data represents the historical business data for the year to which that historical business data belongs.

[0069] The difference in the degree of disorder of the indicator characteristics of the total historical business data and the subset of each conditional attribute reflects the information gain after dividing the total historical business data using each conditional attribute; the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest reflects the degree of synergy between the conditional attribute and the attribute of interest, i.e., net profit, which facilitates the analysis of the impact of the conditional attribute on the company's profitability; the comprehensive analysis of the above two factors yields the initial priority indicators, improving the accuracy of the priority information for decision analysis of each conditional attribute.

[0070] Please see Figure 2 The diagram illustrates a flowchart of an initial priority index acquisition method according to an embodiment of the present invention, the method comprising:

[0071] Step S210: Based on the difference in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subset of each conditional attribute, obtain the information gain of each conditional attribute.

[0072] The degree of disorder of the indicator features of the indicator attributes in the total historical business data reflects the uncertainty before splitting the total historical business data. The degree of disorder of the indicator features of the indicator attributes in the subset of each conditional attribute presents the uncertainty after splitting the total historical business data using each conditional attribute. The difference between the two represents the degree of uncertainty reduction, and the information gain of each conditional attribute is obtained.

[0073] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining information gain includes: obtaining the information entropy of the indicator features of the indicator attributes in the total historical business data, denoted as the total information entropy; obtaining the information entropy of the indicator features of the indicator attributes in the historical business data of each subset of each conditional attribute, denoted as the sub-information entropy of the corresponding subset; taking the proportion of the annual quantity in the historical business data of each subset of each conditional attribute to the annual quantity in the total historical business data as the quantity index of each subset of each conditional attribute; obtaining the overall uncertainty index of each conditional attribute based on the sub-information entropy and quantity index of all subsets of each conditional attribute; and normalizing the difference between the total information entropy and the overall uncertainty index to obtain the information gain of each conditional attribute.

[0074] In this embodiment, the indicator attribute is profitability, which includes two indicator features: yes and no. The probability of the "yes" indicator feature of the indicator attribute in the historical business data of the indicator attribute is... The probability of the "No" indicator feature in the historical business data of indicator attributes is: The total information entropy is determined using the two probabilities mentioned above. The total information entropy represents the uncertainty of total historical business data before decision analysis. The method for calculating the information entropy is well-known to those skilled in the art and will not be elaborated upon here.

[0075] The quantitative index reflects the amount of data in the subset of each conditional attribute. The larger the sub-information entropy and quantitative index of each subset of a conditional attribute, the richer and more complex the information contained in the dataset, and thus the greater the uncertainty of that subset. In this embodiment of the invention, the method for obtaining the overall uncertainty index is as follows: the product of the sub-information entropy and the quantitative index of each subset of each conditional attribute is used as the local uncertainty index of the corresponding subset; the sum of the local uncertainty indices of all subsets of each conditional attribute is used as the overall uncertainty index of each conditional attribute. The larger the overall uncertainty index, the greater the overall level of uncertainty after splitting the total historical business data using each conditional attribute.

[0076] In one specific implementation of this invention, the information gain of each conditional attribute is expressed by the formula:

[0077]

[0078] In the formula, K is the information gain of each conditional attribute; G is the total information entropy; H is the total number of subsets of each conditional attribute. Since this embodiment uses binary splitting, H is taken as an empirical value of 2; I h The quantity index of the h-th subset for each conditional attribute; Jh The sub-information entropy of the h-th subset for each conditional attribute; I h ×J h The local uncertainty index for the h-th subset of data for each conditional attribute; is the overall uncertainty index for each conditional attribute; Norm is the normalization function.

[0079] It should be noted that the larger the information gain K of each conditional attribute, the higher the similarity of the total historical business data divided by that conditional attribute. The more significant the reduction in uncertainty after splitting the total historical business data using that conditional attribute, the more suitable that conditional attribute is for decision analysis.

[0080] Step S220: Based on the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, obtain the collaborative indicators of each conditional attribute and the attribute of interest.

[0081] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the collaborative indicator includes: designating conditional attributes and attention attributes as analytical attributes; sequentially arranging the historical business data of each analytical attribute to obtain a business data sequence for each analytical attribute; obtaining the second-order difference sequence of the business data sequence of each analytical attribute; calculating the mean of the absolute values ​​of the differences between elements with the same index in the second-order difference sequence of each conditional attribute and attention attribute, as a change difference indicator for each conditional attribute and attention attribute; obtaining the correlation coefficient of the business data sequence of each conditional attribute and attention attribute; and obtaining the collaborative indicator for each conditional attribute and attention attribute based on the correlation coefficient and the change difference indicator, wherein the correlation coefficient and the collaborative indicator are positively correlated, and the change difference indicator and the collaborative indicator are negatively correlated.

[0082] Historical business data for each analytical attribute refers to the data in the column containing that analytical attribute in Table 1, including business data for each analytical attribute across all years. Therefore, in this embodiment of the invention, the historical business data for each analytical attribute is arranged chronologically to obtain a business data sequence for the corresponding analytical attribute.

[0083] The second-order difference sequence of the business data series reflects the changes in historical business data for each conditional attribute. The variation difference index presents the similarity between the changing trends of historical business data for each conditional attribute and those of the attribute of interest. The smaller the variation difference index, the more similar the changing trends of historical business data for each conditional attribute are to those of the attribute of interest, and the greater the synergy between the historical business data of each conditional attribute and the historical business data of the attribute of interest (i.e., net profit). When the correlation coefficient between the business data series of each conditional attribute and the attribute of interest is larger, it indicates that net profit is more significantly affected by the historical business data of each conditional attribute, and the synergy between each conditional attribute and net profit is greater. Therefore, the variation difference index and the synergy index are negatively correlated, while the correlation coefficient and the synergy index are positively correlated.

[0084] In this embodiment of the invention, a negative correlation mapping is performed between the change difference index of each conditional attribute and the attribute of interest. The mapping result is then normalized by multiplying the product of the correlation coefficients of the business data sequences of each conditional attribute and the attribute of interest to obtain the synergistic index of each conditional attribute and the attribute of interest. In this embodiment of the invention, other basic mathematical operations can also be used to construct the correlation coefficients and the relationship between the change difference index and the synergistic index of the business data sequences of each conditional attribute and the attribute of interest; this is not limited or elaborated upon here.

[0085] It should be noted that in this embodiment of the invention, an exponential function with the natural constant as the base is used for negative correlation mapping. Specifically, the inverse of the correlation coefficient of the business data sequence of each conditional attribute and the attribute of interest is first taken, and this inverse is used as the exponent of the exponential function with the natural constant as the base to achieve negative correlation mapping of the correlation coefficient. The Norm function is used for normalization, but other methods of negative correlation mapping and normalization can also be selected, which are not limited here. The correlation coefficient used in this embodiment of the invention is the Pearson correlation coefficient. In other embodiments of the invention, the correlation coefficient can also be the Spearman correlation coefficient, the Kendall rank correlation coefficient, the canonical correlation coefficient, etc. The method for obtaining the Pearson correlation coefficient is a well-known technique to those skilled in the art and will not be described in detail here.

[0086] Step S230: Based on the information gain and synergy index, obtain the initial priority index for each conditional attribute; both the information gain and synergy index are positively correlated with the initial priority index.

[0087] The larger the information gain of each conditional attribute, the more significant the reduction in uncertainty of the total historical business data after partitioning it using that attribute, resulting in a purer subset. Therefore, the higher the priority of using that attribute in decision tree construction. Similarly, the larger the synergy index between each conditional attribute and the attribute of interest, the greater the contribution of that attribute to the insurance company's profitability within the total historical business data, making it more suitable for decision-making and thus giving it higher priority in decision tree construction. Therefore, both information gain and synergy index are positively correlated with the initial priority index. In this embodiment, the product of the synergy index of each conditional attribute and the attribute of interest and the information gain is normalized to obtain the initial priority index for each conditional attribute. The larger the initial priority index of each conditional attribute, the higher its priority in decision tree construction.

[0088] In the embodiments of the present invention, the correlation between the synergy index and the information gain and initial priority index can also be constructed through other basic mathematical operations, which will not be limited or elaborated here.

[0089] It should be noted that in this embodiment of the invention, the Sigmoid function is used to normalize the product of the collaboration index and the information gain. Other normalization methods, such as function transformation, max-min normalization, etc., may also be selected in this embodiment of the invention, and are not limited here.

[0090] Step S3: Adjust the initial priority indicators based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes. Use the adjusted initial priority indicators to construct a decision tree and classify and share the total historical business data.

[0091] Because there are often complex relationships between different attributes in financial data—for example, expense ratio and loss ratio both affect profitability; even with high premium income, if both expense ratio and loss ratio rise simultaneously, net profit may still decrease—the degree of correlation between attributes can affect their priority in the decision tree construction process, leading to biases in the high-value data extracted. To construct better decision trees and extract high-value data, it is necessary to consider the correlation between each conditional attribute and other conditional attributes, and combine this with the influence of each conditional attribute on the attribute of interest. This allows for adjustments to the initial priority indicators, resulting in a final priority indicator for each conditional attribute, making the priority of conditional attributes in the decision tree construction process more reasonable.

[0092] Preferably, in some possible implementations of the embodiments of the present invention, the method for obtaining the final priority index includes: obtaining the influence index of each conditional attribute on the attribute of interest; obtaining the correlation coefficient of each conditional attribute with the business data sequence of other conditional attributes, and taking the mean of the correlation coefficients of each conditional attribute with the business data sequence of other conditional attributes as the overall correlation value of each conditional attribute with other conditional attributes; performing a negative correlation mapping on the overall correlation value, and normalizing the product of the mapping result and the influence index of each conditional attribute to obtain the priority construction coefficient of each conditional attribute; and using the priority construction coefficient to weight the initial priority index to obtain the final priority index of each conditional attribute.

[0093] This embodiment analyzes the influence of each conditional attribute on the attribute of interest using a multiple regression function to obtain the influence index. The steps for obtaining the influence index are as follows:

[0094] Linear analysis is performed on all conditional and attention attributes to construct an initial attention function, which includes several regression coefficients. In one specific implementation of this invention, the initial attention function is expressed by the formula:

[0095]

[0096] In the formula, R represents the historical business data of the monitored attributes; A represents the total number of conditional attributes; N represents the total number of conditional attributes. a This refers to the historical business data for the a-th conditional attribute; Let be the regression coefficient of the a-th conditional attribute. These are regression coefficients, all of which are unknowns.

[0097] It should be noted that, since the conditional attributes in this embodiment include insurance income, loss ratio, expense ratio and auto insurance loss ratio, there are a total of 4 conditional attributes and 5 regression coefficients, in order to determine the specific values ​​of the regression coefficients, it is necessary to ensure that the number of years in the total historical business data is much greater than 5. That is, more than 5 years of historical business data of the conditional attributes are needed to participate in the construction of the error function in order to obtain the specific values ​​of the regression coefficients.

[0098] In some possible implementations of this invention, the method for obtaining the error function includes: substituting historical business data of all conditional attributes and historical business data of the attention attributes for each year into an initial attention function to obtain the attention fit value for each year; and obtaining the error function based on the difference between the historical business data of the attention attributes and the attention fit values ​​for all years. In a specific implementation of this invention, the error function RSS is expressed by the formula:

[0099]

[0100] In the formula, U represents the total number of years in the total historical business data, and R... u Let A be the historical business data of the focus attribute for the u-th year in the total historical business data, i.e., the net profit for the u-th year; A is the total number of conditional attributes; N u.a This refers to the historical business data of the a-th conditional attribute in the u-th year of the total historical business data; Let be the regression coefficient of the a-th conditional attribute. These are the regression coefficients; The fitted value for each year.

[0101] The value of the regression coefficient corresponding to the minimum value of the error function is taken as the optimal value of each regression coefficient in the initial focus function, and the optimal value of the regression coefficient corresponding to each conditional attribute is taken as the influence index of each conditional attribute.

[0102] It should be noted that the larger the optimal value of the regression coefficient of each conditional attribute, i.e. the greater the influence index, the greater the impact of that conditional attribute on the attribute of concern, i.e., net profit. The more important that conditional attribute is, the higher its priority in the process of constructing the decision tree.

[0103] To construct better decision trees, conditional attributes with lower correlation to other conditional attributes may provide unique information when segmenting the dataset, thus improving the splitting effect of the decision tree to some extent. The overall correlation value reflects the overall correlation level between each conditional attribute and other conditional attributes; the lower the overall correlation value of each conditional attribute with other conditional attributes, the weaker the overall correlation, the more suitable it is for decision-making, and the higher its priority in the decision tree construction process. Conversely, the larger the influence index of each conditional attribute, the greater its impact on the focus attribute (net profit), the more important the conditional attribute, and the higher its priority in the decision tree construction process. Therefore, the overall correlation value and the priority construction coefficient are negatively correlated, while the influence index and the priority construction coefficient are positively correlated.

[0104] The initial priority index of each conditional attribute is adjusted using the priority construction coefficient. Specifically, the initial priority index is weighted using the priority construction coefficient. The conditional attribute with the greater information gain on net profit, the greater the synergy with net profit, and the smaller the correlation with other conditional attributes has, has a higher priority in building the decision tree.

[0105] In one specific implementation of this invention, the final priority index is expressed by the formula:

[0106]

[0107] In the formula, U is the final priority index for each conditional attribute; L is the initial priority index for each conditional attribute; M is the impact index of each conditional attribute on the attribute of interest; A is the total number of conditional attributes; T a The correlation coefficient between each conditional attribute and the business data sequence of the remaining a-th conditional attributes excluding that conditional attribute; The overall correlation value between each conditional attribute and other conditional attributes; The coefficients are constructed for each conditional attribute; exp is the exponential function with the natural constant as the base; Norm is the normalization function.

[0108] Arrange all conditional attributes in descending order of their final priority indicators to obtain an initial priority sequence. Select the top few conditional attributes from the initial priority sequence to form a final priority sequence. Based on the order of the conditional attributes in the final priority sequence, construct a decision tree using the Hunt algorithm to obtain the classification results of the total historical business data. Store the obtained classification results in a database. The method of constructing the decision tree using the Hunt algorithm is a well-known technique and will not be elaborated upon here. The database used in this embodiment is a Structured Query Language (SQL) database.

[0109] Since financial data involves multiple dimensions of attributes, too many attributes may lead to the complexity of the decision tree model. For example, net profit and revenue growth rate have a strong correlation. These attribute features will increase the complexity of the model and lead to overfitting. In order to reduce the structural redundancy of the decision tree and make the model easier to interpret, several conditional attributes with larger final priority indicators are selected to construct the decision tree.

[0110] In this embodiment, the selection method for conditional attributes in the final priority sequence is as follows: the final priority index of each conditional attribute is normalized to obtain the normalized priority index of the corresponding conditional attribute; for the conditional attributes in the initial priority sequence, the conditional attributes with normalized priority indices greater than the preset priority threshold are arranged in order to obtain the final priority sequence. It should be noted that the preset priority threshold in this embodiment is an empirical value of 0.5, and the implementer can set it according to the specific situation.

[0111] It should be noted that the conditional attributes in the final priority sequence are also arranged from largest to smallest according to the final priority index; the conditional attribute with the larger the final priority index will be used to make decisions on the total historical business data first during the decision tree construction process, that is, the conditional attribute with the larger the final priority index will have a shorter distance between the corresponding node and the root node in the decision tree.

[0112] The leaf nodes of the decision tree represent the classification results of the total historical business data. The classification results of the total historical business data can be directly stored in a database or shared platform, and other companies can also access the database or platform to achieve the purpose of data sharing.

[0113] Since most companies are primarily concerned with their profitability, the company's financial profitability is used as the leaf node of the decision tree. To uncover the factors contributing to a property insurance company's losses and obtain valuable data, in another embodiment, the number of loss-making years for each leaf node of the decision tree is counted. The conditional attribute on the path from the leaf node with the highest number of loss-making years to the root node of the decision tree is recorded as a negative attribute, and the historical business data for each negative attribute is stored in a database. Negative attributes are conditional attributes that negatively impact the company's profitability. Sharing the historical business data for negative attributes allows other insurance companies to focus on these negative attributes and address their negative impact on the company's profitability in a targeted manner.

[0114] This invention is now complete.

[0115] Example 2:

[0116] This invention proposes a data intelligent sharing system for the construction of a smart sharing system. Please refer to [link / reference]. Figure 3 This diagram illustrates a system architecture of a data intelligent sharing system for building a smart sharing system, provided by an embodiment of the present invention. The system includes:

[0117] The data acquisition module 410 is used to acquire total historical business data, which includes historical business data from several years. It extracts business features from the total historical business data and determines the indicator attributes, attention attributes, and several condition attributes in the business features.

[0118] The priority indicator analysis module 420 is used to split the total historical business data by each conditional attribute to obtain different subsets of each conditional attribute; based on the difference in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subsets of each conditional attribute, as well as the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, the initial priority indicator of each conditional attribute is obtained.

[0119] The decision tree construction module 430 is used to adjust the initial priority indicators based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes. The adjusted initial priority indicators are used to construct a decision tree to classify and share the total historical business data.

[0120] It should be noted that the devices provided in the above embodiments are only illustrative examples of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. Furthermore, the data intelligent sharing system and the data intelligent sharing method provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0121] Example 3:

[0122] Figure 4 This is a schematic diagram of a computer device for building a smart sharing system, provided as an embodiment of the present invention. For example, as shown... Figure 4 As shown, the computer device includes: a memory 501, a processor 502, and a computer program 503 stored in the memory 501 and running on the processor 502. When the processor 502 executes the computer program 503, the computer device can execute any of the aforementioned data intelligent sharing methods for building a smart sharing system.

[0123] Furthermore, this application also protects an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to execute a data intelligent sharing method for building a smart sharing system provided in this application.

[0124] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0125] It should be understood that the device provided in this embodiment is used to execute the above-described data intelligent sharing method for building a smart sharing system, and therefore can achieve the same effect as the above-described implementation method.

[0126] When using integrated units, the device may include a processing module and a storage module. When applied to a workpiece, the processing module can be used to control and manage the workpiece's operations. The storage module can be used to support the execution of program code by the workpiece.

[0127] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits contained in conjunction with the disclosure of this application. The processor may also be a combination of functions that implement computing capabilities, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and microprocessors, etc., and the storage module may be a memory.

[0128] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0129] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0130] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A data intelligent sharing method for building a smart sharing system, characterized in that, The method includes: Obtain total historical business data, which includes historical business data from several years. Extract business characteristics from the total historical business data, and determine the indicator attributes, focus attributes, and several condition attributes among the business characteristics. The business characteristics include insurance revenue, loss ratio, expense ratio, net profit, auto insurance loss ratio, and profitability. The indicator attribute is profitability, the focus attribute is net profit, and the condition attributes include: insurance revenue, loss ratio, expense ratio, and auto insurance loss ratio. The total historical business data is split using each conditional attribute to obtain different subsets for each conditional attribute. Based on the differences in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subsets of each conditional attribute, as well as the similarity of the changing trends of the historical business data of each conditional attribute and the attribute of interest, the initial priority indicators for each conditional attribute are obtained. Based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes, the initial priority indicators are adjusted, and a decision tree is constructed using the adjusted initial priority indicators to classify and share the total historical business data. Based on the difference in the degree of disorder of the indicator characteristics of the indicator attributes in the total historical business data and the subset of each conditional attribute, obtain the information gain of each conditional attribute. Based on the similarity of the changing trends of historical business data for each conditional attribute and the attribute of interest, obtain the collaborative indicators for each conditional attribute and the attribute of interest. Based on information gain and synergy metrics, the initial priority index for each conditional attribute is obtained; both information gain and synergy metrics are positively correlated with the initial priority index. Obtain the collaborative metrics for each conditional attribute and the attribute of interest, including: The conditional attributes and the attributes of interest are recorded as analysis attributes. The historical business data of each analysis attribute are arranged sequentially to obtain the business data sequence of each analysis attribute. Obtain the second-order difference sequence of the business data sequence for each analytical attribute; calculate the mean of the absolute values ​​of the differences between the elements with the same index in the second-order difference sequence of each conditional attribute and the attribute of interest, and use it as an indicator of the difference in change between each conditional attribute and the attribute of interest. Obtain the correlation coefficient of the business data sequence of each conditional attribute and the attribute of interest. Based on the correlation coefficient and the change difference index, obtain the coordination index of each conditional attribute and the attribute of interest. The correlation coefficient and the coordination index are positively correlated, while the change difference index and the coordination index are negatively correlated. The total historical business data is split using each conditional attribute to obtain different subsets for each conditional attribute, including: The product of the maximum value in the historical business data of each conditional attribute and the preset weight threshold is used as the standard segmentation threshold for each conditional attribute. Based on the standard segmentation threshold, the historical business data of each conditional attribute is divided into two categories; the set consisting of the historical business data of each year to which each data belongs in each category of historical business data for each conditional attribute is used as the subset of each conditional attribute.

2. The data intelligent sharing method for building a smart sharing system according to claim 1, characterized in that, Obtain the information gain for each conditional attribute, including: Obtain the information entropy of the indicator characteristics of the indicator attributes in the total historical business data, and denote it as the total information entropy; Obtain the information entropy of the indicator features of the indicator attributes in the historical business data of each subset of each conditional attribute, and denote it as the sub-information entropy of the corresponding subset; take the proportion of the annual quantity in the historical business data of each subset of each conditional attribute to the annual quantity in the total historical business data as the quantity indicator of each subset of each conditional attribute. Based on the sub-information entropy and quantity index of all subsets of each conditional attribute, obtain the overall uncertainty index of each conditional attribute; The difference between the total information entropy and the overall uncertainty index is normalized to obtain the information gain of each conditional attribute.

3. The data intelligent sharing method for building a smart sharing system according to claim 1, characterized in that, Based on the influence of each conditional attribute on the attribute of interest and the correlation between each conditional attribute and other conditional attributes, the initial priority indicators are adjusted. A decision tree is then constructed using the adjusted initial priority indicators. The total historical business data is then classified and shared, including: Based on the degree of influence of each conditional attribute on the attribute of interest and the degree of correlation between each conditional attribute and other conditional attributes, the initial priority indicators are adjusted to obtain the final priority indicator for each conditional attribute. Arrange all conditional attributes in descending order of their final priority to obtain an initial priority sequence; select the first few conditional attributes from the initial priority sequence to form the final priority sequence. Based on the order of the conditional attributes in the final priority sequence, a decision tree is constructed using the Hunt algorithm to obtain the classification results of the total historical business data, and the obtained classification results are stored in the database.

4. The data intelligent sharing method for building a smart sharing system according to claim 3, characterized in that, Obtain the final priority metric for each conditional attribute, including: Obtain the impact metrics of each conditional attribute on the attribute of interest; Obtain the correlation coefficient of each conditional attribute with the business data sequence of other conditional attributes, and take the average of the correlation coefficients of each conditional attribute with the business data sequence of other conditional attributes as the overall correlation value of each conditional attribute with other conditional attributes. A negative correlation mapping is performed on the overall correlation values, and the product of the mapping result and the influence index of each conditional attribute is normalized to obtain the priority construction coefficient of each conditional attribute. The initial priority index is weighted using priority construction coefficients to obtain the final priority index for each conditional attribute.

5. A data intelligent sharing method for building a smart sharing system according to claim 4, characterized in that, Obtain the impact metrics of each conditional attribute on the attribute of interest, including: Linear analysis is performed on all conditional and interest attributes to construct an initial interest function, which contains several regression coefficients. Substitute the historical business data of all conditional attributes and the historical business data of the attention attributes for each year into the initial attention function to obtain the attention fit value for each year. Based on the differences between historical business data and fitted values ​​of the focus attributes for all years, an error function is obtained; the value of the regression coefficient corresponding to the minimum value of the error function is taken as the optimal value of each regression coefficient in the initial focus function, and the optimal value of the regression coefficient corresponding to each condition attribute is taken as the influence index of each condition attribute.

6. A data intelligent sharing method for building a smart sharing system according to claim 2, characterized in that, Based on the sub-information entropy and quantity indices of all subsets of each conditional attribute, obtain the overall uncertainty index for each conditional attribute, including: The product of the sub-information entropy and the quantity index of each subset of each conditional attribute is used as the local uncertainty index of the corresponding subset; the sum of the local uncertainty indices of all subsets of each conditional attribute is used as the overall uncertainty index of each conditional attribute.

7. A data intelligent sharing method for building a smart sharing system according to claim 2, characterized in that, The correlation coefficient is the Pearson correlation coefficient.

Images