Dynamic index management system and method based on data layering

Abstract

The invention discloses a dynamic index management system and method based on data layering, and particularly relates to the technical field of accumulation fund management, and the method comprises the steps: 1, obtaining the change data of a target user, and carrying out the analysis based on the change data, so as to obtain a change evaluation coefficient of the target user; 2, performing comparative analysis on the basis of the change evaluation coefficient and a preset evaluation coefficient threshold, and obtaining n low-frequency data and i high-frequency data according to a comparison result; 3, acquiring change characteristic data of the target user in a historical time period based on the high-frequency data; the high-frequency data and the low-frequency data are respectively processed through data layering, so that the processing efficiency of the system can be remarkably improved; the feature prediction model is used for predicting the future change feature data, and the change reason is determined based on the similarity, so that misjudgment can be effectively avoided, and the accuracy of change reason analysis is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of provident fund management, and more specifically, to a dynamic indicator management system and method based on data stratification. Background Art

[0002] Data stratification refers to the process of dividing a data set into several levels or categories according to certain standards and rules. The purpose of data stratification is to better organize, manage and analyze data in order to extract useful information from it. Data at different levels can have different granularity and characteristics, and can provide different perspectives and insights at all levels. When analyzing housing provident funds, the use of data stratification is an effective way to analyze, organize and manage data. Housing provident fund dynamic indicators refer to a set of variables or data related to housing provident funds, which can change over time and reflect the status and trend of housing provident funds. These indicators include but are not limited to: number of depositors, deposit amount, withdrawal amount, loan balance, overdue rate, etc. Through the monitoring and analysis of dynamic indicators, the operation status of housing provident funds can be understood in a timely manner, and effective management and decision-making can be carried out. Traditional housing provident fund dynamic indicators usually rely on manual data entry and periodic report generation, which cannot reflect the dynamic changes of housing provident funds in a timely manner, so there is a delay, which in turn affects the timeliness and effectiveness of decision-making.

[0003] In the prior art, the data in the database is analyzed according to the bank's indicators and the user's personal information, and the amount of loan, loan interest, and housing provident fund withdrawal amount of the user are analyzed, which reduces the workload, avoids data errors and calculation errors, and improves work efficiency. For example, the Chinese patent application with publication number CN109166028A discloses a business system. Although the above method can improve work efficiency to a certain extent, the inventor has studied and applied the above method and the prior art and found that the above method and the prior art have at least the following defects:

[0004] 1. Due to the limitations of existing data analysis methods and tools, it is impossible to fully process and analyze a large amount of provident fund data, resulting in errors and anomalies in the analysis results; these errors and anomalies make the results of provident fund indicator analysis unreliable, thus affecting the accuracy of decision-making;

[0005] 2. Existing electronic auditing tools are unable to accurately determine the specific reasons for changes in provident funds after each inspection, resulting in frequent misjudgments; misjudgments increase the complexity of subsequent management and processing, and reduce management efficiency and quality.

[0006] To this end, the present invention provides a dynamic indicator management system and method based on data stratification. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art, the present invention provides a dynamic indicator management system and method based on data stratification to solve the problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, in a first aspect, the present invention provides the following technical solutions: a dynamic indicator management method based on data stratification, comprising:

[0009] Step 1: Obtain the target user's change data, and analyze based on the change data to obtain the target user's change evaluation coefficient;

[0010] Step 2: Perform comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtain n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero;

[0011] Step 3: Obtain the target user's change feature data within the historical time period based on high-frequency data;

[0012] Step 4: Input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T;

[0013] Step 5: Determine whether the provident fund indicator of the target user changes in the future time period T according to the indicator dynamic coefficient. If there is no change, set T=T+a and return to step 1; if there is a change, determine the associated change indicator information based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero;

[0014] Step 6: Generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal.

[0015] Furthermore, the change assessment coefficient includes a user information change assessment coefficient, a deposit record change assessment coefficient, a withdrawal record change assessment coefficient, and a loan record change assessment coefficient;

[0016] Methods for analyzing change data to obtain the change evaluation coefficient of target users include:

[0017] Step a1: Obtain change data within a preset time period according to the timestamp;

[0018] Specifically, the change data includes user information change value, deposit record change data value, withdrawal record change data value and loan record change data value;

[0019] Step a2: obtaining the change standard value of each target user, wherein the change standard value includes the user information change standard value, the deposit record change standard value, the withdrawal record change standard value and the loan record change standard value;

[0020] Step a3: Compare the user information change standard value with the user information change value to calculate the user information change evaluation coefficient; the calculation formula is: Where: Yhp represents the user information change evaluation coefficient, Yhx represents the user information change value, Yhx0 represents the user information change standard value, ln(·) represents the logarithmic function with e as the base;

[0021] Step a4: Compare the deposit record change standard value with the deposit record change data value to calculate the deposit record change assessment coefficient; the calculation formula is: Where: Jcp represents the deposit record change assessment coefficient, Jcb represents the deposit record change data value, and Jcb0 represents the deposit record change standard value;

[0022] Step a5: Compare the extraction record change standard value with the extraction record change data value to calculate the extraction record change assessment coefficient; the calculation formula is: Where: Tqp represents the extraction record change assessment coefficient, Tqb represents the extraction record change data value, Tqb0 represents the extraction record change standard value, ln(·) represents the logarithmic function with e as the base;

[0023] Step a6: Compare the loan record change standard value with the loan record change data value to calculate the loan record change assessment coefficient; the calculation formula is: In the formula: Dkp represents the loan record change assessment coefficient, Dkb represents the loan record change data value, Dkb0 represents the loan record change standard value, and e represents a natural constant greater than zero.

[0024] Furthermore, the comparison and analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold is performed to obtain n low-frequency data and i high-frequency data according to the comparison result, including:

[0025] Preset the evaluation coefficient threshold, and compare the changed evaluation coefficient with the preset evaluation coefficient threshold;

[0026] If the change evaluation coefficient is less than the preset evaluation coefficient threshold, the corresponding change evaluation coefficient is marked as low-frequency data;

[0027] If the change evaluation coefficient is greater than or equal to the preset evaluation coefficient threshold, the corresponding change evaluation coefficient is marked as high-frequency data;

[0028] The preset evaluation coefficient thresholds include a first evaluation coefficient threshold, a second evaluation coefficient threshold, a third evaluation coefficient threshold and a fourth evaluation coefficient threshold; the change evaluation coefficient and the preset evaluation coefficient threshold are replaced by one of the user information change evaluation coefficient and the first evaluation coefficient threshold, the deposit record change evaluation coefficient in the change evaluation coefficient and the second evaluation coefficient threshold, the extraction record change evaluation coefficient in the change evaluation coefficient and the third evaluation coefficient threshold, or the loan record change evaluation coefficient in the change evaluation coefficient and the fourth evaluation coefficient threshold, to obtain n low-frequency data and i high-frequency data respectively.

[0029] Furthermore, the change characteristic data includes a deposit ratio difference set, an extraction frequency difference set and a loan repayment rate difference set of the target user.

[0030] Further, the preconfigured feature prediction models include M, where M is an integer greater than zero; it should be further noted that the M feature prediction models include but are not limited to a feature prediction model for predicting a future deposit ratio mean value within a future time period T, a feature prediction model for predicting a future extraction frequency mean value within a future time period T, and a feature prediction model for predicting a future loan repayment rate mean value within a future time period T;

[0031] The future change characteristic data include the future average of the deposit ratio, the future average of the withdrawal frequency and the future average of the loan repayment rate;

[0032] The generation logic of the feature prediction model is as follows:

[0033] Acquire historical deposit ratio sample data, and divide the historical deposit ratio sample data into a future deposit ratio training set and a future deposit ratio test set; the historical deposit ratio sample data includes a deposit ratio difference set and its corresponding future deposit ratio mean value;

[0034] Construct a regression network model, use the difference set of future deposit ratios in the future deposit ratio training set as input data of the regression network model, use the mean of future deposit ratios in the future deposit ratio training set as output data of the regression network model, train the regression network model, and obtain an initial feature prediction network;

[0035] The initial feature prediction network is model-verified using the future deposit ratio test set, and the initial feature prediction network with a prediction error less than or equal to a preset error threshold is output as the feature prediction model;

[0036] The regression network model is one of the models such as LSTM neural network, decision tree regression network, random forest regression network, support vector machine regression network or linear regression network.

[0037] Furthermore, the corresponding indicator dynamic coefficient is calculated according to the future change characteristic data of the target user in the future time period T, including:

[0038] The future deposit ratio mean, future withdrawal frequency mean and future loan repayment rate mean in the future change characteristic data of the target user in the future time period T are marked as Wj, Wt and Wd respectively;

[0039] The future average deposit ratio, future average withdrawal frequency and future average loan repayment rate are dimensionless and then calculated using a formula to obtain the indicator dynamic coefficient. The calculation formula is as follows:

[0040]

[0041] Where: Zp represents the dynamic coefficient of the indicator, β1 is the weight factor of the future mean deposit ratio, β2 is the weight factor of the future mean withdrawal frequency, β3 is the weight factor of the future mean loan repayment rate, C is a correction constant greater than zero, and ln(·) represents the logarithmic function with e as the base.

[0042] Furthermore, it is determined whether there is a change in the provident fund indicator of the target user in the future time period T, including:

[0043] Step b1. Compare the indicator dynamic coefficient with H indicator dynamic coefficient intervals, where H is an integer greater than zero;

[0044] Step b2. If the indicator dynamic coefficient falls into any one of the H indicator dynamic coefficient intervals, it is determined that there is a change in the provident fund indicator of the target user in the future time period T;

[0045] Step b3. If the indicator dynamic coefficient does not fall into any of the H indicator dynamic coefficient intervals, it is determined that there is no change in the target user in the future time period T.

[0046] Furthermore, there are F actual change performance graphs, and the F actual change performance graphs include an actual change performance graph reflecting changes in the deposit ratio, an actual change performance graph reflecting changes in the withdrawal frequency, or an actual change performance graph reflecting changes in the loan repayment rate;

[0047] Generate an actual change performance graph based on the future change feature data in the future time period T, including:

[0048] Extract each deposit ratio value of future change characteristic data within the future time period T;

[0049] A two-dimensional coordinate system is constructed, with the future time period T as the horizontal axis of the two-dimensional coordinate system and each contribution ratio value as the vertical axis of the two-dimensional coordinate system, to obtain an actual change performance graph reflecting the change in the contribution ratio.

[0050] Furthermore, the final change reason of the target user is determined according to the similarity between the actual change performance graph and the standard change performance graph, including:

[0051] Step c1. Calculate the similarity of the same form of the F actual change performance diagrams and the F standard change performance diagrams using a similarity algorithm to obtain W similarities, where W is an integer greater than zero;

[0052] It should be noted that the similarity algorithm includes but is not limited to the cosine similarity algorithm or the Euclidean distance algorithm, etc.; the actual change performance graph and the standard change performance graph need to be preprocessed before comparison, and the preprocessing includes but is not limited to image enhancement, image denoising and image segmentation, etc.

[0053] Step c2. Filter the maximum similarity among the W similarities, and compare the maximum similarity with the preset similarity;

[0054] Step c3. If the maximum similarity is greater than or equal to the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is used as the final change reason;

[0055] Step c4. If the maximum similarity is less than the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is excluded as the final change reason.

[0056] In a second aspect, the present invention provides a dynamic indicator management system based on data stratification, which is used to implement the above-mentioned dynamic indicator management method based on data stratification, comprising:

[0057] A data analysis module is used to obtain the change data of the target user and analyze the change data to obtain the change evaluation coefficient of the target user;

[0058] The coefficient comparison module performs comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtains n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero;

[0059] The feature data acquisition module acquires the target user's changing feature data within a historical period based on high-frequency data;

[0060] The prediction module is used to input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T;

[0061] A judgment module is used to judge whether the provident fund indicator of the target user has changed in the future time period T according to the indicator dynamic coefficient. If there is no change, T=T+a is set, and the process returns to step 1; if there is a change, the associated change indicator information is determined based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero;

[0062] The management module is used to generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal.

[0063] Technical effects and advantages of the present invention:

[0064] 1. The present invention processes high-frequency data and low-frequency data separately through data stratification, which can significantly improve the processing efficiency of the system; the feature prediction model is used to predict future change feature data, and the cause of the change is determined based on the similarity, which can effectively avoid misjudgment and improve the accuracy of the analysis of the cause of the change; dynamic monitoring and prediction are performed based on historical data, and the change feature data in the future time period is predicted by the feature prediction model, and the dynamic coefficient of the indicator is calculated, which can predict the change of the provident fund indicator in advance and help the management department take corresponding measures.

[0065] 2. The present invention conducts a detailed analysis of the target user's provident fund change data and generates a corresponding change performance chart. The system can provide detailed feedback on the reasons for the change to the provident fund management terminal. The feedback mechanism can help the management department better understand user behavior and needs, thereby improving decision support, optimizing provident fund management strategies, and improving the overall efficiency and quality of housing provident fund management. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A flow chart of a dynamic indicator management method based on data stratification in Example 1;

[0067] Figure 2 This is a structural diagram of a dynamic indicator management system based on data stratification according to Example 2;

[0068] Figure 3 This is a schematic diagram of an electronic device according to Embodiment 3;

[0069] Figure 4 A schematic diagram of a computer-readable storage medium of Example 4. DETAILED DESCRIPTION

[0070] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0072] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and a similar second unit may be referred to as a first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0073] Example 1

[0074] See also Figure 1 As shown, this embodiment discloses a dynamic indicator management method based on data stratification, including:

[0075] Step 1: Obtain the target user's change data, and analyze based on the change data to obtain the target user's change evaluation coefficient;

[0076] It should be noted that: the change data is used to characterize the changes in housing provident funds; the change assessment coefficients include user information change assessment coefficients, payment record change assessment coefficients, withdrawal record change assessment coefficients and loan record change assessment coefficients.

[0077] Methods for analyzing change data to obtain the change evaluation coefficient of target users include:

[0078] Step a1: Obtain change data within a preset time period according to the timestamp;

[0079] Specifically, the change data includes user information change values, deposit record change data values, withdrawal record change data values ​​and loan record change data values.

[0080] It should be noted that the user information change value, deposit record change data value, withdrawal record change data value and loan record change data value are the specific values ​​or changes when changes occur in each data, such as changes in the target user's contact information or changes in the deposit base; it should be supplemented that the user information includes name, ID number, contact information, work unit and position, etc.; the deposit record includes deposit date, deposit amount and deposit ratio, etc.; the withdrawal record includes withdrawal date, withdrawal amount and withdrawal type (buying a house or renting a house), etc.; the loan record includes loan application date, loan amount, loan term, repayment date and repayment amount, etc.

[0081] Step a2: obtaining the change standard value of each target user, wherein the change standard value includes the user information change standard value, the deposit record change standard value, the withdrawal record change standard value and the loan record change standard value;

[0082] Step a3: Compare the user information change standard value with the user information change value to calculate the user information change evaluation coefficient; the calculation formula is: Where: Yhp represents the user information change evaluation coefficient, Yhx represents the user information change value, Yhx0 represents the user information change standard value, ln(·) represents the logarithmic function with e as the base;

[0083] Step a4: Compare the deposit record change standard value with the deposit record change data value to calculate the deposit record change assessment coefficient; the calculation formula is: Where: Jcp represents the deposit record change assessment coefficient, Jcb represents the deposit record change data value, and Jcb0 represents the deposit record change standard value;

[0084] Step a5: Compare the extraction record change standard value with the extraction record change data value to calculate the extraction record change assessment coefficient; the calculation formula is: Where: Tqp represents the extraction record change assessment coefficient, Tqb represents the extraction record change data value, Tqb0 represents the extraction record change standard value, ln(·) represents the logarithmic function with e as the base;

[0085] Step a6: Compare the loan record change standard value with the loan record change data value to calculate the loan record change assessment coefficient; the calculation formula is: In the formula: Dkp represents the loan record change assessment coefficient, Dkb represents the loan record change data value, Dkb0 represents the loan record change standard value, and e represents a natural constant greater than zero.

[0086] Step 2: Perform comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtain n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero;

[0087] In a specific implementation, the comparison analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold is performed, and n low-frequency data and i high-frequency data are obtained according to the comparison result, including:

[0088] Preset an evaluation coefficient threshold, compare the change evaluation coefficient with the preset evaluation coefficient threshold, and the preset evaluation coefficient threshold is determined by those skilled in the art based on actual needs and a large amount of experimental data;

[0089] If the change evaluation coefficient is less than the preset evaluation coefficient threshold, the corresponding change evaluation coefficient is marked as low-frequency data, indicating that the corresponding change evaluation coefficient belongs to low-frequency data, and the low-frequency data is stored in a low-cost storage method, and the low-cost storage method is a traditional relational database such as MySQL, PostgreSQL or Amazon S3;

[0090] If the change assessment coefficient is greater than or equal to the preset assessment coefficient threshold, the corresponding change assessment coefficient will be marked as high-frequency data, indicating that the corresponding change assessment coefficient belongs to high-frequency data. The high-frequency data will be quickly read, written and processed using a high-performance database, and the high-performance database is a NoSQL database such as Redis, MongoDB or Cassandra.

[0091] It should be noted that: the preset evaluation coefficient threshold includes a first evaluation coefficient threshold, a second evaluation coefficient threshold, a third evaluation coefficient threshold and a fourth evaluation coefficient threshold; in a specific implementation, the change evaluation coefficient and the preset evaluation coefficient threshold can be replaced by one of the user information change evaluation coefficient and the first evaluation coefficient threshold, the deposit record change evaluation coefficient in the change evaluation coefficient and the second evaluation coefficient threshold, the extraction record change evaluation coefficient in the change evaluation coefficient and the third evaluation coefficient threshold, or the loan record change evaluation coefficient in the change evaluation coefficient and the fourth evaluation coefficient threshold, to obtain n low-frequency data and i high-frequency data respectively.

[0092] Step 3: Obtain the target user's change feature data within the historical time period based on high-frequency data;

[0093] Specifically, the change characteristic data includes a deposit ratio difference set, a withdrawal frequency difference set, and a loan repayment rate difference set of the target user.

[0094] The contribution ratio is the proportion of the target user's salary used to contribute to the housing provident fund; this embodiment takes the individual contribution ratio as an example; the withdrawal frequency is the number of times the target user withdraws the housing provident fund within a certain period of time, reflecting the target user's demand for the use of the provident fund; the loan repayment rate is the proportion of the target user's timely repayment of the housing provident fund loan, usually expressed as a percentage.

[0095] It should be noted that: the deposit ratio difference set includes multiple deposit ratio differences within the preset monitoring time period; the preset time period includes R historical time periods, R is an integer greater than zero; a further example is: assuming that the historical time period is from January 1 to December 31, and the preset monitoring time period is December, the deposit ratio difference within the preset monitoring time period includes the deposit ratio difference between January 1 and January 31, the deposit ratio difference between February 1 and February 28, the deposit ratio difference between March 1 and March 31,..., the deposit ratio difference between December 1 and December 31; similarly, the same is true for the extraction frequency difference set and the loan repayment rate difference set. For details, please refer to the introduction of the deposit ratio difference set, and no further elaboration will be given.

[0096] Step 4: Input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T;

[0097] Specifically, the preconfigured feature prediction models include M, where M is an integer greater than zero; it should be further noted that the M feature prediction models include but are not limited to a feature prediction model for predicting a future deposit ratio mean value within a future time period T, a feature prediction model for predicting a future extraction frequency mean value within a future time period T, and a feature prediction model for predicting a future loan repayment rate mean value within a future time period T;

[0098] Specifically, the future change characteristic data include the future average of the deposit ratio, the future average of the withdrawal frequency, and the future average of the loan repayment rate;

[0099] In a specific implementation, the generation logic of the feature prediction model is as follows:

[0100] Acquire historical deposit ratio sample data, and divide the historical deposit ratio sample data into a future deposit ratio training set and a future deposit ratio test set; the historical deposit ratio sample data includes a deposit ratio difference set and its corresponding future deposit ratio mean value;

[0101] Construct a regression network model, use the difference set of future deposit ratios in the future deposit ratio training set as input data of the regression network model, use the mean of future deposit ratios in the future deposit ratio training set as output data of the regression network model, train the regression network model, and obtain an initial feature prediction network;

[0102] The initial feature prediction network is model-verified using the future deposit ratio test set, and the initial feature prediction network with a prediction error less than or equal to a preset error threshold is output as the feature prediction model;

[0103] It should be noted that: the regression network model is one of the models such as LSTM neural network, decision tree regression network, random forest regression network, support vector machine regression network or linear regression network;

[0104] It should also be noted that: the generation logic of the above-mentioned feature prediction model is specifically the training process of a specific feature prediction model among the M feature prediction models, which is specifically the training process of a feature prediction model for predicting the future average deposit ratio in the future time period T, wherein the future average deposit ratio is obtained by adding up the monthly deposit ratios and dividing it by the time length N (the calculation logic of other future variable feature data is the same, which will not be elaborated on); but the generation logic of the above-mentioned feature prediction model is also the same as the generation logic of the remaining M-1 feature prediction models, so it will not be elaborated on, and the details can be referred to the above-mentioned training process; it should be noted that: the M feature prediction models only have different prediction results, and their training principles and processes are the same.

[0105] In implementation, the corresponding indicator dynamic coefficient is calculated based on the future change characteristic data of the target user in the future time period T, including:

[0106] The future deposit ratio mean, future withdrawal frequency mean and future loan repayment rate mean in the future change characteristic data of the target user in the future time period T are marked as Wj, Wt and Wd respectively;

[0107] The future average deposit ratio, future average withdrawal frequency and future average loan repayment rate are dimensionless and then calculated using a formula to obtain the indicator dynamic coefficient. The calculation formula is as follows:

[0108]

[0109] Wherein: Zp represents the dynamic coefficient of the indicator, β1 is the weight factor of the future mean deposit ratio, β2 is the weight factor of the future mean extraction frequency, β3 is the weight factor of the future mean loan repayment rate, and the weight factors are all set by technical personnel in this field based on experience, C is a correction constant greater than zero, and ln(·) represents a logarithmic function with e as the base.

[0110] Step 5: Determine whether the provident fund indicator of the target user changes in the future time period T according to the indicator dynamic coefficient. If there is no change, set T=T+a and return to step 1; if there is a change, determine the associated change indicator information based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero;

[0111] It should be noted that a can be a natural month or set by relevant staff.

[0112] In implementation, it is determined whether there are changes in the provident fund indicators of the target users in the future time period T, including:

[0113] Step b1. Compare the indicator dynamic coefficient with H indicator dynamic coefficient intervals, where H is an integer greater than zero;

[0114] Step b2. If the indicator dynamic coefficient falls into any one of the H indicator dynamic coefficient intervals, it is determined that there is a change in the provident fund indicator of the target user in the future time period T;

[0115] Step b3. If the indicator dynamic coefficient does not fall into any of the H indicator dynamic coefficient intervals, it is determined that there is no change in the target user in the future time period T.

[0116] It should be noted that: H indicator dynamic coefficient intervals are pre-stored in the system database, and the specific range of each indicator dynamic coefficient interval is determined according to the experimental situation. Each indicator dynamic coefficient interval is specifically and artificially associated with a change indicator information in advance. In one change indicator information, there are G change reasons and F standard change performance graphs corresponding to each change reason. The F standard change performance graphs include a standard change performance graph reflecting changes in deposit ratio, a standard change performance graph reflecting changes in withdrawal frequency, or a standard change performance graph reflecting changes in loan repayment rate.

[0117] It should be understood that: since the preset relationship between the indicator dynamic coefficient interval and the change indicator information is established through a pre-set binding; therefore, when the indicator dynamic coefficient is compared with H indicator dynamic coefficient intervals, the indicator dynamic coefficient interval to which the indicator dynamic coefficient corresponds can be obtained, and then according to the mapping relationship between the indicator dynamic coefficient interval and the change indicator information, the corresponding change indicator information mapped by the indicator dynamic coefficient interval to which the indicator dynamic coefficient corresponds can be known, that is, the associated change indicator information is determined; further, the G reasons for the change when the provident fund indicator of the target user is about to change in the future time period T and the F standard change performance diagrams corresponding to each change reason are known, and then according to subsequent processing, the specific situation is further determined to achieve quick investigation in advance, which is conducive to shortening the provident fund indicator management time of the target user and improving the provident fund management effect.

[0118] Step 6: Generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal;

[0119] In a specific implementation, generating an actual change performance graph according to future change characteristic data in a future time period T includes:

[0120] Extract each deposit ratio value of future change characteristic data within the future time period T;

[0121] A two-dimensional coordinate system is constructed, with the future time period T as the horizontal axis of the two-dimensional coordinate system and each contribution ratio value as the vertical axis of the two-dimensional coordinate system, to obtain an actual change performance graph reflecting the change in the contribution ratio.

[0122] It should be noted that: there are F actual change performance graphs, and in other specific implementations, the vertical axis of the two-dimensional coordinate system can be replaced by each extraction frequency of future change characteristic data in the future time period T or each loan repayment rate of future change characteristic data in the future time period T; therefore, the F actual change performance graphs include but are not limited to actual change performance graphs reflecting changes in the deposit ratio, actual change performance graphs reflecting changes in the extraction frequency, or actual change performance graphs reflecting changes in the loan repayment rate; the generation logic of the actual change performance graphs in other forms is the same as the generation logic of the actual change performance graphs reflecting changes in the deposit ratio, and will not be described in detail;

[0123] In implementation, the final change reason of the target user is determined based on the similarity between the actual change performance graph and the standard change performance graph, including:

[0124] Step c1. Calculate the similarity of the same form of the F actual change performance diagrams and the F standard change performance diagrams using a similarity algorithm to obtain W similarities, where W is an integer greater than zero;

[0125] It should be noted that the similarity algorithm includes but is not limited to the cosine similarity algorithm or the Euclidean distance algorithm, etc.; the actual change performance graph and the standard change performance graph need to be preprocessed before comparison, and the preprocessing includes but is not limited to image enhancement, image denoising and image segmentation, etc.

[0126] Step c2. Filter the maximum similarity among the W similarities, and compare the maximum similarity with the preset similarity;

[0127] Step c3. If the maximum similarity is greater than or equal to the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is used as the final change reason;

[0128] Step c4. If the maximum similarity is less than the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is excluded as the final change reason;

[0129] It should be understood that: when the change cause corresponding to the standard change performance diagram with the maximum similarity is excluded as the final change cause, the remaining G-1 change causes are further checked until the maximum similarity of the standard change performance diagram associated with a certain change cause is greater than or equal to the preset similarity, and the final change cause is obtained.

[0130] This embodiment processes high-frequency data and low-frequency data separately through data stratification, which can significantly improve the processing efficiency of the system; uses the feature prediction model to predict future change feature data and determines the cause of the change based on similarity, which can effectively avoid misjudgment and improve the accuracy of the analysis of the cause of the change; performs dynamic monitoring and prediction based on historical data, predicts the change feature data in the future time period through the feature prediction model, and calculates the dynamic coefficient of the indicator, which can predict the change of the provident fund indicator in advance and help the management department take corresponding measures.

[0131] In this embodiment, by conducting a detailed analysis of the target user's provident fund change data and generating a corresponding change performance chart, the system can provide the provident fund management terminal with detailed feedback on the reasons for the changes. The feedback mechanism can help the management department better understand user behavior and needs, thereby improving decision support, optimizing provident fund management strategies, and improving the overall efficiency and quality of housing provident fund management.

[0132] Example 2

[0133] See also Figure 2 As shown, this embodiment provides a dynamic indicator management system based on data stratification, including:

[0134] A data analysis module is used to obtain the change data of the target user and analyze the change data to obtain the change evaluation coefficient of the target user;

[0135] The coefficient comparison module performs comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtains n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero;

[0136] The feature data acquisition module acquires the target user's changing feature data within a historical period based on high-frequency data;

[0137] The prediction module is used to input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T;

[0138] A judgment module is used to judge whether the provident fund indicator of the target user has changed in the future time period T according to the indicator dynamic coefficient. If there is no change, T=T+a is set, and the process returns to step 1; if there is a change, the associated change indicator information is determined based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero;

[0139] The management module is used to generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal.

[0140] Example 3

[0141] See also Figure 3 As shown, this embodiment provides an electronic device 500. The electronic device 500 may include one or more processors and one or more memories. The memories store computer readable codes, and when the computer readable codes are executed by one or more processors, a dynamic indicator management method based on data stratification as described above may be executed.

[0142] The system according to the embodiment of the present application can also be used Figure 3 The electronic device architecture shown in FIG. Figure 3As shown, the electronic device 500 may include a bus 501, one or more CPUs 502, a read-only memory (ROM) 503, a random access memory (RAM) 504, a communication port 505 connected to a network, an input / output 506, a hard disk 507, etc. The storage device in the electronic device 500, such as the ROM 503 or the hard disk 507, may store a dynamic indicator management method based on data stratification provided by the present application. Furthermore, the electronic device 500 may also include a user interface 508. Of course, Figure 3 The architecture shown is only exemplary and can be omitted according to actual needs when implementing different devices. Figure 3 One or more components of an electronic device are shown.

[0143] Example 4

[0144] See also Figure 4 As shown, this embodiment provides a computer-readable storage medium storing instructions. When the instructions are executed on a computer, the computer executes a dynamic indicator management method based on data stratification in Embodiment 1.

[0145] The formulas involved in the above are all calculated by removing dimensions and taking their numerical values. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The weight factors in the formula and the preset thresholds in the analysis process are set by technicians in this field according to actual conditions or obtained by simulating a large amount of data; the size of the weight factor is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the weight factor depends on the amount of sample data and the corresponding processing coefficient initially set by technicians in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0146] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0147] Finally: 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 spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A dynamic indicator management method based on data stratification, characterized in that: include: Step 1: Obtain the target user's change data, and analyze based on the change data to obtain the target user's change evaluation coefficient; Step 2: Perform comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtain n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero; Step 3: Obtain the target user's change feature data within the historical time period based on high-frequency data; Step 4: Input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T; Step 5: Determine whether the provident fund indicator of the target user changes in the future time period T according to the indicator dynamic coefficient. If there is no change, set T=T+a and return to step 1; if there is a change, determine the associated change indicator information based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero; Step 6: Generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal.

2. According to claim 1, a dynamic indicator management method based on data stratification is characterized in that: The change assessment coefficients include user information change assessment coefficients, deposit record change assessment coefficients, withdrawal record change assessment coefficients and loan record change assessment coefficients; Methods for analyzing the change data to obtain the change evaluation coefficient of the target user include: Step a1: Obtain change data within a preset time period according to the timestamp; The change data includes user information change value, deposit record change data value, withdrawal record change data value and loan record change data value; Step a2: obtaining the change standard value of each target user, wherein the change standard value includes the user information change standard value, the deposit record change standard value, the withdrawal record change standard value and the loan record change standard value; Step a3: Compare the user information change standard value with the user information change value to calculate the user information change evaluation coefficient; the calculation formula is: Where: Yhp represents the user information change evaluation coefficient, Yhx represents the user information change value, Yhx0 represents the user information change standard value, ln(·) represents the logarithmic function with e as the base; Step a4: Compare the deposit record change standard value with the deposit record change data value to calculate the deposit record change assessment coefficient; the calculation formula is: Where: Jcp represents the deposit record change assessment coefficient, Jcb represents the deposit record change data value, and Jcb0 represents the deposit record change standard value; Step a5: Compare the extraction record change standard value with the extraction record change data value to calculate the extraction record change assessment coefficient; the calculation formula is: Where: Tqp represents the extraction record change assessment coefficient, Tqb represents the extraction record change data value, Tqb0 represents the extraction record change standard value, ln(·) represents the logarithmic function with e as the base; Step a6: Compare the loan record change standard value with the loan record change data value to calculate the loan record change assessment coefficient; the calculation formula is: In the formula: Dkp represents the loan record change assessment coefficient, Dkb represents the loan record change data value, Dkb0 represents the loan record change standard value, and e represents a natural constant greater than zero.

3. A dynamic index management method based on data stratification according to claim 2, characterized in that: The comparison and analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold is performed, and n low-frequency data and i high-frequency data are obtained according to the comparison result, including: Preset the evaluation coefficient threshold, and compare the changed evaluation coefficient with the preset evaluation coefficient threshold; If the change evaluation coefficient is less than the preset evaluation coefficient threshold, the corresponding change evaluation coefficient is marked as low-frequency data; If the change evaluation coefficient is greater than or equal to the preset evaluation coefficient threshold, the corresponding change evaluation coefficient is marked as high-frequency data; The preset evaluation coefficient thresholds include a first evaluation coefficient threshold, a second evaluation coefficient threshold, a third evaluation coefficient threshold and a fourth evaluation coefficient threshold; the change evaluation coefficient and the preset evaluation coefficient threshold are replaced by one of the user information change evaluation coefficient and the first evaluation coefficient threshold, the deposit record change evaluation coefficient in the change evaluation coefficient and the second evaluation coefficient threshold, the extraction record change evaluation coefficient in the change evaluation coefficient and the third evaluation coefficient threshold, or the loan record change evaluation coefficient in the change evaluation coefficient and the fourth evaluation coefficient threshold, to obtain n low-frequency data and i high-frequency data respectively.

4. A dynamic index management method based on data stratification according to claim 3, characterized in that: The change characteristic data includes a deposit ratio difference set, an extraction frequency difference set and a loan repayment rate difference set of the target user.

5. A dynamic index management method based on data stratification according to claim 4, characterized in that: The preconfigured feature prediction models include M, where M is an integer greater than zero; the M feature prediction models include but are not limited to a feature prediction model for predicting the mean value of the future deposit ratio within the future time period T, a feature prediction model for predicting the mean value of the future extraction frequency within the future time period T, and a feature prediction model for predicting the mean value of the future loan repayment rate within the future time period T; The future change characteristic data include the future average of the deposit ratio, the future average of the withdrawal frequency and the future average of the loan repayment rate; The generation logic of the feature prediction model is as follows: Acquire historical deposit ratio sample data, and divide the historical deposit ratio sample data into a future deposit ratio training set and a future deposit ratio test set; the historical deposit ratio sample data includes a deposit ratio difference set and its corresponding future deposit ratio mean value; Construct a regression network model, use the difference set of future deposit ratios in the future deposit ratio training set as input data of the regression network model, use the mean of future deposit ratios in the future deposit ratio training set as output data of the regression network model, train the regression network model, and obtain an initial feature prediction network; The initial feature prediction network is model-verified using the future deposit ratio test set, and the initial feature prediction network with a prediction error less than or equal to a preset error threshold is output as the feature prediction model; The regression network model is one of the models such as LSTM neural network, decision tree regression network, random forest regression network, support vector machine regression network or linear regression network.

6. A dynamic index management method based on data stratification according to claim 5, characterized in that: The corresponding indicator dynamic coefficient is calculated based on the future change characteristic data of the target user in the future time period T, including: The future deposit ratio mean, future withdrawal frequency mean and future loan repayment rate mean in the future change characteristic data of the target user in the future time period T are marked as Wj, Wt and Wd respectively; The future average deposit ratio, future average withdrawal frequency and future average loan repayment rate are dimensionless and then calculated using a formula to obtain the indicator dynamic coefficient. The calculation formula is as follows: Where: Zp represents the dynamic coefficient of the indicator, β1 is the weight factor of the future mean deposit ratio, β2 is the weight factor of the future mean withdrawal frequency, β3 is the weight factor of the future mean loan repayment rate, C is a correction constant greater than zero, and ln(·) represents the logarithmic function with e as the base.

7. A dynamic index management method based on data stratification according to claim 6, characterized in that: Determine whether there are changes in the provident fund indicators of the target user in the future time period T, including: Step b1. Compare the indicator dynamic coefficient with H indicator dynamic coefficient intervals, where H is an integer greater than zero; Step b2. If the indicator dynamic coefficient falls into any one of the H indicator dynamic coefficient intervals, it is determined that there is a change in the provident fund indicator of the target user in the future time period T; Step b3. If the indicator dynamic coefficient does not fall into any of the H indicator dynamic coefficient intervals, it is determined that there is no change in the target user in the future time period T.

8. A dynamic index management method based on data stratification according to claim 7, characterized in that: There are F actual change performance graphs, and the F actual change performance graphs include an actual change performance graph reflecting changes in the deposit ratio, an actual change performance graph reflecting changes in the withdrawal frequency, or an actual change performance graph reflecting changes in the loan repayment rate; Generate an actual change performance graph based on the future change feature data in the future time period T, including: Extract each deposit ratio value of future change characteristic data within the future time period T; A two-dimensional coordinate system is constructed, with the future time period T as the horizontal axis of the two-dimensional coordinate system and each contribution ratio value as the vertical axis of the two-dimensional coordinate system, to obtain an actual change performance graph reflecting the change in the contribution ratio.

9. The method for dynamic index management based on data stratification according to claim 8, characterized in that: Determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, including: Step c1. Calculate the similarity of the same form of the F actual change performance diagrams and the F standard change performance diagrams using a similarity algorithm to obtain W similarities, where W is an integer greater than zero; Step c2. Filter the maximum similarity among the W similarities, and compare the maximum similarity with the preset similarity; Step c3. If the maximum similarity is greater than or equal to the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is used as the final change reason; Step c4. If the maximum similarity is less than the preset similarity, the change reason corresponding to the standard change performance diagram with the maximum similarity is excluded as the final change reason.

10. A dynamic indicator management system based on data stratification, used to implement a dynamic indicator management method based on data stratification according to any one of claims 1 to 9, characterized in that: include: A data analysis module is used to obtain the change data of the target user and analyze the change data to obtain the change evaluation coefficient of the target user; The coefficient comparison module performs comparative analysis based on the change evaluation coefficient and the preset evaluation coefficient threshold, and obtains n low-frequency data and i high-frequency data according to the comparison result, where n and i are both integers greater than zero; The feature data acquisition module acquires the target user's changing feature data within a historical period based on high-frequency data; The prediction module is used to input the change feature data in the historical time period into the pre-configured corresponding feature prediction model to obtain the future change feature data of the target user in the future time period T; and calculate the corresponding indicator dynamic coefficient based on the future change feature data of the target user in the future time period T; A judgment module is used to judge whether the provident fund indicator of the target user has changed in the future time period T according to the indicator dynamic coefficient. If there is no change, T=T+a is set, and the process returns to step 1; if there is a change, the associated change indicator information is determined based on the preset relationship between the indicator dynamic coefficient interval and the change indicator information; the change indicator information includes G change reasons and F standard change performance diagrams corresponding to each change reason, where G, F and a are integers greater than zero; The management module is used to generate an actual change performance graph based on the future change feature data in the future time period T, determine the final change reason of the target user based on the similarity between the actual change performance graph and the standard change performance graph, and feed back the change reason to the provident fund management terminal.

Images