A method and system for early warning monitoring of enterprise operations based on data analysis

Abstract

The present invention discloses a method and system for monitoring enterprise operation early warning based on data analysis, which relates to the technical field of data analysis. The method comprises the following steps: acquiring enterprise data to establish an operation prediction model, collecting operation stability information in the operation prediction model, analyzing the state of the operation prediction model, classifying and marking the operation prediction model into different states according to the analysis result, and then analyzing the state transition of the classified and marked operation prediction model, generating a risk early warning index according to the transition details information of the business data under the transition state, comparing the risk early warning index with the data risk threshold, generating a risk data signal and a standard recommendation signal, and performing a joint early warning in combination with the state of the operation prediction model, thereby determining the separation between the business data and the model and improving the accuracy of the early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a method and system for early warning monitoring of enterprise operations based on data analysis. Background Art

[0002] Enterprise operation early warning is a management method that aims to detect potential problems, risks or opportunities early by monitoring key business indicators and other relevant data in real time or near real time, so as to take timely and effective measures to deal with these situations. Enterprise operation early warning helps to improve the agility of the enterprise, reduce operational risks, optimize resource utilization, and better adapt to the ever-changing market environment. Data analysis plays a key role in enterprise operation early warning. By collecting, processing, analyzing and interpreting large amounts of data inside and outside the enterprise, valuable information and insights can be extracted to identify potential problems and opportunities, support decision-making, and provide early warning information in real time or near real time.

[0003] The existing technology has the following deficiencies: the existing enterprise operation risks are generally analyzed through the internal data of the enterprise, but the enterprise is not only affected by the internal production, human resources, inventory levels and other related operations, but also by the external market competition in the same industry and the supply chain. The enterprise's operation warning is judged only by the internal influence of the enterprise, and the actual operation risk of the enterprise cannot be analyzed from multiple data. The reliability of the enterprise operation warning is reduced and there is a lack of support for the warning.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for early warning monitoring of enterprise operations based on data analysis to address the deficiencies in the background technology.

[0006] In order to achieve the above-mentioned object, the present invention provides the following technical solutions: a method for early warning monitoring of enterprise operations based on data analysis, comprising the following steps;

[0007] Acquire enterprise data to establish an operation forecast model, collect operational stability information in the operation forecast model, analyze the status of the operation forecast model, and classify and mark the operation forecast model into different statuses based on the analysis results;

[0008] Analyze the state transition of the classification and tagging operation prediction model, and generate a risk warning index based on the transition details of the business data under the transition state;

[0009] Compare the risk warning index with the data risk threshold to generate risk data signals and standard recommendation signals, and conduct joint warnings based on the status of the operational prediction model.

[0010] Preferably, the operational stability information in the operational forecasting model is collected to determine the status of the operational forecasting model, specifically including: the operational stability information includes trend fluctuation information and analytical stability information, the trend fluctuation information includes the sales-supply overlap coefficient and is calibrated as XSG, and the analytical stability information includes the monitoring error deviation value and is calibrated as JCB.

[0011] Preferably, the logic for obtaining the sales supply overlap coefficient and the monitoring error deviation value is as follows:

[0012] Get the actual sales value data set SJ={SJj}={SJ1, SJ2, …, SJy} and the corresponding predicted sales value data set YQ={YQj}={YQ1, YQ2, …, YQy} within a period of time. For each time point, calculate the difference value DY=|SJ-YQ| between the actual sales and the predicted sales, and establish the difference value set: DY={DYj}={DY1, DY2, …, DYy}, where y is a positive integer. Calculate the difference value, actual sales value, and predicted sales value to obtain the sales supply overlap coefficient. The calculation expression is:

[0013] Get the actual number of errors in each monitoring aspect of the enterprise per unit time t and the expected number of errors set, take the absolute value of the difference between the actual number of errors in each monitoring aspect and the predicted number of errors as the monitoring residual value and establish a monitoring residual value set: CZ = {CZi} = {CZ1, CZ2, ..., CZx}, x is a positive integer, x represents the monitoring aspect, calculate the error value of the number of errors: RMS = ∑CZi 2 / x, establish a set of error values, RMSz = {RMS1, RMS2, ..., RMSz}, z is a positive integer, and calculate the mean error value Calculate the monitoring error deviation value, the calculation expression is:

[0014] Preferably, the acquired sales supply overlap coefficient and monitoring error deviation value are used to generate an accurate analysis index;

[0015] The sales-supply overlap coefficient and monitoring error deviation value are directly proportional to the accurate analysis index.

[0016] Preferably, the operation prediction model is classified and marked into different states according to the analysis results. The specific process is as follows:

[0017] The generated accuracy analysis index is compared with the first accuracy assessment threshold and the second accuracy assessment threshold, the first accuracy assessment threshold is greater than the second accuracy assessment threshold, and the state of the operation prediction model is classified according to the comparison result;

[0018] If the accuracy analysis index is less than or equal to the first accuracy assessment threshold and greater than the second accuracy assessment threshold, the operation prediction model state is marked as an abnormal state;

[0019] If the accuracy analysis index is greater than the first accuracy assessment threshold, the operation prediction model state is marked as normal;

[0020] If the accuracy analysis index is less than or equal to the second accuracy assessment threshold, the operation prediction model state is marked as a fault state.

[0021] Preferably, the risk warning index is generated according to the change details information of the business data in the change state, and the specific process is as follows:

[0022] Collect the change details when the operation forecast model state changes. The change details include feature selection information and demand change information. The feature selection information includes the factor feature selection coefficient. The demand change information includes the processing efficiency floating value. After collection, the factor feature selection coefficient and the processing efficiency floating value are calibrated as YST and SCX respectively.

[0023] Obtain the enterprise operation risk monitoring related data within a unit time, obtain the number of changes CS of the operation prediction model, obtain the change data set of the operation prediction model, obtain the number of risk categories TS of the change data set, and calculate the purity of the feature data. The calculation formula is: c is the number of categories, p r Indicates the proportion of the data of the rth category in the transition data set to the total data set. Perform purity analysis on the transition nodes of each transition number, establish a feature purity set GDu = {GD1, GD2, ..., GDw}, w is a positive integer, and calculate the feature purity mean Calculate the factor feature selection coefficient, the calculation expression is:

[0024] Obtain the business data volume for relevant historical risk business analysis and establish a business data volume set, obtain the time required for business analysis and establish a business analysis time set, take the quotient of business data volume and business analysis time as business efficiency, establish a business efficiency set XLk = {XL1, XL2, ..., XLd}, d is a positive integer, and obtain the mean of the business efficiency set Get the real-time business processing efficiency SXL of the model state transition. The calculation formula of the processing efficiency floating value is:

[0025] Preferably, the obtained factor feature selection coefficient and processing efficiency floating value are dimensionless, and the risk warning index is established after removing the unit, and calibrated as F z , based on the formula: In the formula, e1 and e2 are the factor feature selection coefficients LB d And the area purchase floating value GM f The preset proportional coefficient, and e1 and e1 are both greater than 0.

[0026] Preferably, the risk warning index is compared with the data risk threshold to generate a risk data signal and a standard recommendation signal. The specific steps are as follows:

[0027] Compare the risk warning index with the data risk threshold;

[0028] If the risk warning index is less than the data risk threshold, a standard data signal is generated for the analyzed business data;

[0029] If the risk warning index is greater than or equal to the data risk threshold, a risk data signal is generated for the analyzed business data.

[0030] Preferably, a joint warning is performed in combination with the status of the operation prediction model, and the specific steps are as follows:

[0031] Confirm the business data that generates risk data signals and analyze the relationship between the operational forecast model status and business data;

[0032] When the state of the operation prediction model changes from abnormal state and fault state to normal state, re-analyze the business data and contact relevant maintenance personnel to process the business data;

[0033] When the state of the operational prediction model changes from a normal state to an abnormal state and a fault state, a corresponding category warning is issued for the category of the business data that generates the risk data signal, and the enterprise is notified to adjust its strategy;

[0034] When the business data generates a normal data signal and the state of the operation prediction model changes from a normal state to an abnormal state and a fault state, a model warning is issued to the operation prediction model.

[0035] The present invention also provides an enterprise operation early warning monitoring based on data analysis, including a data acquisition module, a data processing module, and a data early warning module;

[0036] Acquire enterprise data to establish an operation prediction model, collect operation stability information in the operation prediction model, analyze the status of the operation prediction model, classify and mark the operation prediction model into different states according to the analysis results, and send the collected data to the data processing module;

[0037] The data processing module receives the data sent by the data acquisition module, analyzes the state transition of the classification and labeling operation prediction model, generates a risk warning index based on the transition details of the business data under the transition state, and sends it to the data warning module;

[0038] The data warning module receives the data sent by the data processing module, compares the risk warning index with the data risk threshold, generates risk data signals and standard recommendation signals, and conducts joint warnings based on the status of the operation prediction model.

[0039] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0040] The present application analyzes the operational stability information of the enterprise's operational forecasting model, combines the trend fluctuation information in the operational stability information with the analytical stability information, generates an accurate analysis index, compares the generated accurate analysis index with the accurate assessment first threshold and the accurate assessment second threshold, classifies the state of the operational forecasting model according to the comparison result, obtains the model state according to the analysis, thereby determining the accuracy of the model and the reference of the risk warning, and then analyzes the business data during the state change, thereby determining a specific warning based on the analysis result, generates a risk warning index by collecting feature selection information of the state change of the operational forecasting model and combining it with the demand change information, compares the risk warning index with the data risk threshold, generates different data signals according to the comparison result, records the generated data signal as a risk data signal in real time, and realizes different warnings in combination with the operational forecasting model state, thereby determining the separation between the business data and the model, thereby improving the accuracy of the warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0042] Figure 1 The present invention is a flowchart of a method and system for early warning monitoring of enterprise operations based on data analysis.

[0043] Figure 2 The present invention is a structural schematic diagram of an enterprise operation early warning monitoring method and system based on data analysis. DETAILED DESCRIPTION

[0044] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0045] Embodiment 1: The present invention provides Figure 1 A method for early warning monitoring of enterprise operations based on data analysis is shown, the method comprising the following steps:

[0046] Acquire enterprise data to establish an operation forecast model, collect operational stability information in the operation forecast model, and determine the status of the operation forecast model;

[0047] Analyze the status of the operation prediction model and classify and mark the operation prediction model into different statuses according to the analysis results;

[0048] Analyze the state transition of the classified and marked operation prediction model, and generate a risk warning index based on the transition details of the business data in the transition state;

[0049] Compare the risk warning index with the data risk threshold to generate risk data signals and standard recommendation signals, and conduct joint warnings in combination with the status of the operational prediction model.

[0050] Before generating an operational forecasting model, ensure that the business problem to be solved is clearly defined, the forecasting goal is clear, and how the model should help the business, and obtain data related to the business problem, including historical business data, market data, customer data, competitor data, etc., to ensure the quality, integrity and legality of the data; conduct preliminary exploratory data analysis (EDA) on the data to understand the distribution, correlation, missingness, etc. of the data, and handle missing values, outliers and duplicate values ​​in the data.

[0051] Perform preprocessing steps such as data normalization, encoding categorical variables, and processing time series to prepare data for modeling; divide the data set into training sets and test sets to ensure that the model has good generalization capabilities on unseen data; select an appropriate model based on the business problem and data characteristics, and try multiple models. For example, the linear regression model is used to predict a continuous target variable and compare their performance for problems with linear relationships, such as house price prediction and sales volume prediction. The random forest model is used for classification and regression problems, and is particularly suitable for large-scale, high-dimensional data. The specific model is selected based on the actual situation.

[0052] In business operations, external warnings and internal warnings usually influence and relate to each other. The generation of operational warnings depends on different situations and business environments. External warnings affect internal warnings. For example, interruptions, delays or other problems in the external supply chain may directly affect internal production and supply chain warnings. Problems such as scarcity or price fluctuations of production raw materials will lead to rising internal costs or production interruptions. Internal warnings affect external warnings. For example, problems with internal production efficiency lead to delayed product delivery, which in turn causes dissatisfaction among external customers, affects the company's reputation, and causes external warnings for the company.

[0053] Collect the company's operational stability information, which includes trend fluctuation information and analytical stability information. The collected information is used to analyze the company's accuracy when making subsequent market forecasts.

[0054] Trend fluctuation information includes the sales-supply overlap coefficient, and analysis stability information includes the monitoring error deviation value. After collection, the sales-supply overlap coefficient and the monitoring error deviation value are calibrated as XSG and JCB respectively.

[0055] The sales-supply overlap coefficient in the trend fluctuation information plays an important role in analyzing the early warning of enterprise operations. The sales-supply overlap coefficient indicates the overlap between the actual operational supply changes of the enterprise and the predicted supply changes. The larger the sales-supply overlap coefficient, the more consistent or similar the changes between the actual sales and the predicted sales. The enterprise can more accurately predict market demand, and procurement and production are relatively more reasonable. The sales-supply overlap coefficient has the following effects on the accuracy of analysis and prediction:

[0056] Evaluate forecast accuracy: The sales-supply overlap coefficient is an indicator that measures the similarity between actual sales and forecasted sales. A higher overlap coefficient indicates that the forecast is relatively accurate and consistent with actual sales. It is used to intuitively evaluate the accuracy of supply chain forecasts.

[0057] Provide real-time feedback: The sales-supply overlap coefficient can be used to monitor the consistency between forecasts and actual sales in real time. If the overlap coefficient decreases, it means that the forecast accuracy has decreased, and it is necessary to adjust the supply chain strategy or improve the forecast model in time to better adapt to market changes;

[0058] Enhance supply chain agility: By continuously monitoring the sales-supply overlap coefficient, companies can respond more swiftly to market changes. Highly accurate forecasts support more flexible supply chain planning, helping to reduce risks and improve business adaptability.

[0059] The logic for obtaining the sales supply overlap coefficient is as follows:

[0060] Get the actual sales value data set SJ={SJj}={SJ1, SJ2, …, SJy} and the corresponding predicted sales value data set YQ={YQj}={YQ1, YQ2, …, YQy} within a period of time. For each time point, calculate the difference value DY=|SJ-YQ| between the actual sales and the predicted sales, and establish the difference value set: DY={DYj}={DY1, DY2, …, DYy}, where y is a positive integer. Calculate the difference value, actual sales value, and predicted sales value to obtain the sales supply overlap coefficient. The calculation expression is:

[0061] It should be noted that a period of time is set according to actual conditions, such as a month, quarter, etc. The time span and frequency of data element collection in the actual sales value dataset and the corresponding predicted sales value dataset are consistent.

[0062] The monitoring error deviation value in the analysis stability information plays an important role in the accuracy of the analysis and forecast. The monitoring error deviation value indicates the error deviation between the enterprise operation forecast and the actual situation in the process of market trend forecasting. The monitoring error deviation value will have the following impacts:

[0063] Decision quality: The existence of monitoring error deviation values ​​can affect the quality of corporate decision-making. If the monitoring error deviation value is large, management may make wrong decisions based on inaccurate forecasts, leading to problems such as improper resource allocation, overproduction or insufficient inventory;

[0064] Enterprise competitiveness: The error of monitoring error deviation value affects the competitiveness of enterprises in the market. Accurate prediction helps to flexibly adjust market strategies, launch new products and better meet customer needs, thereby improving the competitiveness of enterprises;

[0065] The logic for obtaining the monitoring error deviation value is as follows:

[0066] Get the actual number of errors in each monitoring aspect of the enterprise per unit time t and the expected number of errors set, take the absolute value of the difference between the actual number of errors in each monitoring aspect and the predicted number of errors as the monitoring residual value and establish a monitoring residual value set: CZ = {CZi} = {CZ1, CZ2, ..., CZx}, x is a positive integer, x represents the monitoring aspect, calculate the error value of the number of errors: RMS = ∑CZi 2 / x, establish a set of error values, RMSz = {RMS1, RMS2, ..., RMSz}, z is a positive integer, and calculate the mean error value Calculate the monitoring error deviation value, the calculation expression is:

[0067] It should be noted that there are multiple time points for collecting data within a unit of time. If corresponding data is collected at different time points, the specific time points are set according to actual conditions.

[0068] Comprehensively analyze the accuracy of enterprise operation forecasts based on trend fluctuation information and analysis stability information, and obtain the accuracy of the operation forecast model based on the analysis results;

[0069] The obtained sales supply overlap coefficient and monitoring error deviation value are processed dimensionlessly, and the accurate analysis index is generated after removing the unit, and calibrated as P z , based on the formula: Wherein, α and γ are the preset proportional coefficients of the sales supply overlap coefficient XSG and the monitoring error deviation value JCB, respectively, and α and γ are greater than 0.

[0070] From the formula, we can see that the larger the sales supply overlap coefficient, the smaller the monitoring error deviation value, that is, the accuracy analysis index P z The larger the performance value, the higher the accuracy analysis index, indicating that the accuracy of the operation forecast model is higher, the smaller the sales supply overlap coefficient, and the larger the monitoring error deviation value, that is, the accuracy analysis index P z The smaller the performance value, the lower the accuracy of the operation forecasting model.

[0071] By comparing the accuracy analysis index with the preset accuracy threshold, the operation forecast model can be divided into different states. According to the state splitting results, the analysis of the operation forecast model can be more deeply understood, thereby optimizing the enterprise operation strategy;

[0072] The generated accuracy analysis index is compared with the first accuracy assessment threshold and the second accuracy assessment threshold, the first accuracy assessment threshold is greater than the second accuracy assessment threshold, and the state of the operation prediction model is classified according to the comparison result;

[0073] After obtaining the generated accurate analysis index, the generated accurate analysis index is compared with the first accurate assessment threshold and the second accurate assessment threshold respectively. If the accurate analysis index is less than or equal to the first accurate assessment threshold and greater than the second accurate assessment threshold, the operation prediction model status is marked as abnormal, indicating that the operation prediction model is prone to enterprise operation warning errors when performing data analysis, and more time and comparison are needed to make accurate warning analysis before using it for enterprise operation;

[0074] If the accuracy analysis index is greater than the first threshold of accuracy assessment, the operation prediction model status is marked as normal, indicating that when the operation prediction model performs enterprise operation analysis, the analysis result is consistent with the actual enterprise warning, and the operation prediction model can be used for normal analysis;

[0075] If the accuracy analysis index is less than or equal to the second accuracy assessment threshold, the operation prediction model status is marked as a faulty status, indicating that the enterprise operation analysis results of the operation prediction model are seriously deviated from the actual situation, and the operation prediction model has a fault and cannot be used normally.

[0076] The present application analyzes the operational stability information of the enterprise's operational forecasting model, combines the trend fluctuation information in the operational stability information with the analytical stability information, generates an accurate analysis index, compares the generated accurate analysis index with the accurate assessment first threshold and the accurate assessment second threshold, classifies the status of the operational forecasting model according to the comparison results, and obtains the model status according to the analysis, thereby determining the accuracy of the model and the reference value of the risk warning.

[0077] Example 2: The state of the operation forecast model often changes between states within a continuous time period. Possible reasons for the change of the state of the operation forecast model include but are not limited to data quality issues, insufficient model complexity, improper feature selection, market changes, future information leakage, etc. Analyzing the detailed reasons helps to identify the root cause of the problem and provide guidance for model optimization. It also analyzes the factors that cause the state of the operation forecast model to deteriorate, and then determines whether to focus on related enterprise operation early warning.

[0078] Collect the change details information when the operation prediction model state changes. The change details information includes feature selection information and demand change information. The feature selection information includes the factor feature selection coefficient. The demand change information includes the processing efficiency floating value. After collection, the factor feature selection coefficient and the processing efficiency floating value are calibrated as YST and SCX respectively.

[0079] The factor feature selection coefficient is used to analyze the state fluctuations of the operation forecast model transformation process. The factor feature selection coefficient indicates the accuracy of extracting relevant risks and identifying corresponding risk features when monitoring enterprise operation risks. A large factor feature selection coefficient indicates that there is a large difference in features when selecting features, which may easily lead to a large deviation between data when the operation forecast model is analyzed, thereby generating an early warning. The factor feature selection coefficient has the following effects:

[0080] Real-time monitoring indicators: Based on the extracted key features, enterprises can establish real-time monitoring indicators to monitor changes in these features. Once these features change abnormally, it indicates potential operational risks, thus triggering corresponding response measures;

[0081] Importance assessment: The factor feature selection coefficient provides an indicator to assess the importance of each feature to the enterprise's operational risk. By analyzing the size and direction of the factor feature selection coefficient, the enterprise can identify which factors have a more significant impact on the risk and help understand the main driving factors of the risk.

[0082] The logic for obtaining the factor feature selection coefficient is as follows:

[0083] Obtain the enterprise operation risk monitoring related data within a unit time, obtain the number of changes CS of the operation prediction model, obtain the change data set of the operation prediction model, obtain the number of risk categories TS of the change data set, and calculate the purity of the feature data. The calculation formula is: c is the number of categories, p r Indicates the proportion of the data of the rth category in the transition data set to the total data set. Perform purity analysis on the transition nodes of each transition number, establish a feature purity set GDu = {GD1, GD2, ..., GDw}, w is a positive integer, and calculate the feature purity mean Calculate the factor feature selection coefficient, the calculation expression is:

[0084] It should be noted that enterprise operational risk-related data include financial data, market data, supply chain data, customer feedback, etc. The risk data belonging to each field are classified to obtain the number of risk categories. The transition data set refers to the data set analyzed when the state of the operational prediction model is changed.

[0085] The processing efficiency float value indicates the fluctuation or change in the efficiency of processing related business data. The processing efficiency float value can help enterprises evaluate the changes in their operational efficiency. By monitoring the efficiency, enterprises can find bottlenecks and problems in the business process and take timely measures to adjust and improve them. By analyzing the float value, enterprises can formulate strategies and decisions more accurately to adapt to the changing operating environment.

[0086] The logic for obtaining the floating value of processing efficiency is as follows:

[0087] Obtain the business data volume for relevant historical risk business analysis and establish a business data volume set, obtain the time required for business analysis and establish a business analysis time set, take the quotient of business data volume and business analysis time as business efficiency, establish a business efficiency set XLk = {XL1, XL2, ..., XLd}, d is a positive integer, and obtain the mean of the business efficiency set Get the real-time business processing efficiency SXL of the model state transition. The calculation formula of the processing efficiency floating value is:

[0088] It should be noted that relevant risk business refers to different business data, such as financial data, market data, supply chain data and other analytical data. Different business data need to be classified accordingly when analyzing. When establishing a set, the business efficiency is first compared with the set threshold. After determining that it meets the standards, the business efficiency is placed in the business efficiency set, and data that does not meet the standards is pre-screened and eliminated.

[0089] The obtained factor feature selection coefficient and processing efficiency floating value are dimensionless, and the risk warning index is established after removing the unit, and calibrated as F z , based on the formula: In the formula, e1 and e2 are the factor feature selection coefficients LB d And the area purchase floating value GM f The preset proportional coefficient, and e1 and e2 are both greater than 0.

[0090] From the formula, we can see that the larger the factor feature selection coefficient, the larger the processing efficiency floating value, that is, the risk warning index F z The larger the performance value, the worse the data selection of the operation prediction model is when making risk predictions. The smaller the factor feature selection coefficient is, the smaller the processing efficiency floating value is, that is, the risk warning index F z The smaller the performance value, the better the data selection of the operation prediction model is when making risk predictions, and the data has stronger characteristics.

[0091] Compare the risk warning index with the data risk threshold. The specific process is as follows:

[0092] If the risk warning index is less than the data risk threshold, a standard data signal is generated for the analyzed business data, indicating that the data selection of the operation prediction model is normal when performing relevant data business analysis;

[0093] If the risk warning index is greater than or equal to the data risk threshold, a risk data signal is generated for the analyzed business data, indicating that the data does not meet the requirements when the operation prediction model performs relevant data business analysis, resulting in more errors in the processing process.

[0094] Confirm the business data that generates risk data signals and analyze the relationship between the operational prediction model status and business data. The specific process is as follows:

[0095] When the state of the operation prediction model changes from abnormal state and fault state to normal state, re-analyze the business data and contact relevant maintenance personnel to process the business data;

[0096] When the state of the operation prediction model changes from a normal state to an abnormal state and a fault state, a corresponding category warning is issued for the category of the business data that generates the risk data signal, and the enterprise is notified to adjust its strategy.

[0097] When the business data generates a normal data signal, when the state of the operation prediction model changes from a normal state to an abnormal state and a fault state, a model warning is issued to the operation prediction model.

[0098] This application analyzes the business data when the state changes, so as to determine the specific warning according to the analysis results, and generates a risk warning index by collecting the feature selection information of the state change of the operation prediction model and combining it with the demand change information. The risk warning index is compared with the data risk threshold, and different data signals are generated according to the comparison results. The generated data signals are recorded in real time as risk data signals, and different warnings are realized in combination with the state of the operation prediction model, thereby determining the separation between the business data and the model and improving the accuracy of the warning.

[0099] Embodiment 3: The present invention provides Figure 2 The enterprise operation early warning monitoring system based on data analysis shown includes a data acquisition module, a data processing module, and a data early warning module;

[0100] Acquire enterprise data to establish an operation prediction model, collect operation stability information in the operation prediction model, analyze the status of the operation prediction model, classify and mark the operation prediction model into different states according to the analysis results, and send the collected data to the data processing module;

[0101] The data processing module receives the data sent by the data acquisition module, analyzes the state transition of the classification and labeling operation prediction model, generates a risk warning index based on the transition details of the business data under the transition state, and sends it to the data warning module;

[0102] The data warning module receives the data sent by the data processing module, compares the risk warning index with the data risk threshold, generates risk data signals and standard recommendation signals, and conducts joint warnings based on the status of the operation prediction model.

[0103] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0104] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0105] It should be noted that, in this article, if there are relational terms such as first and second, etc., they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0106] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0107] 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.

Claims

1. A method for early warning and monitoring of enterprise operations based on data analysis, characterized in that: The steps include: Acquire enterprise data to establish an operation forecast model, collect operational stability information in the operation forecast model, analyze the status of the operation forecast model, and classify and mark the operation forecast model into different statuses based on the analysis results; Analyze the state transition of the classification and tagging operation prediction model, and generate a risk warning index based on the transition details of the business data under the transition state; Compare the risk warning index with the data risk threshold to generate risk data signals and standard recommendation signals, and conduct joint warnings in combination with the status of the operational prediction model; Collect the operational stability information in the operational forecast model and determine the status of the operational forecast model, including: operational stability information includes trend fluctuation information and analytical stability information. Trend fluctuation information includes sales supply overlap coefficient and is calibrated as XSG. Analytical stability information includes monitoring error deviation value and is calibrated as JCB. The logic for obtaining the sales supply overlap coefficient and monitoring error deviation value is as follows: Get the actual sales value data set SJ={SJj}={SJ1, SJ2, …, SJy} and the corresponding predicted sales value data set YQ={YQj}={YQ1, YQ2, …, YQy} within a period of time. For each time point, calculate the difference value DY=|SJ-YQ| between the actual sales and the predicted sales, and establish the difference value set: DY={DYj}={DY1, DY2, …, DYy}, where y is a positive integer. Calculate the difference value, actual sales value, and predicted sales value to obtain the sales supply overlap coefficient. The calculation expression is: Get the actual number of errors in each monitoring aspect of the enterprise per unit time t and the expected number of errors set, take the absolute value of the difference between the actual number of errors in each monitoring aspect and the predicted number of errors as the monitoring residual value and establish a monitoring residual value set: CZ = {CZi} = {CZ1, CZ2, ..., CZx}, x is a positive integer, x represents the monitoring aspect, calculate the error value of the number of errors: RMS = ∑CZi 2 / x, establish a set of error values, RMSz = {RMS1, RMS2, ..., RMSz}, z is a positive integer, and calculate the mean error value Calculate the monitoring error deviation value, the calculation expression is: Generate accurate analysis index using the acquired sales supply overlap coefficient and monitoring error deviation value; The sales-supply overlap coefficient and monitoring error deviation value are directly proportional to the accurate analysis index.

2. According to the enterprise operation early warning monitoring method based on data analysis according to claim 1, it is characterized in that: According to the analysis results, the operation prediction model is classified and marked into different states. The specific process is as follows: The generated accuracy analysis index is compared with the first accuracy assessment threshold and the second accuracy assessment threshold, the first accuracy assessment threshold is greater than the second accuracy assessment threshold, and the state of the operation prediction model is classified according to the comparison result; If the accuracy analysis index is less than or equal to the first accuracy assessment threshold and greater than the second accuracy assessment threshold, the operation prediction model state is marked as an abnormal state; If the accuracy analysis index is greater than the first accuracy assessment threshold, the operation prediction model state is marked as normal; If the accuracy analysis index is less than or equal to the second accuracy assessment threshold, the operation prediction model state is marked as a fault state.

3. The enterprise operation early warning monitoring method based on data analysis according to claim 2 is characterized in that: Generate a risk warning index based on the change details of the business data in the change state. The specific process is as follows: Collect the change details when the operation forecast model state changes. The change details include feature selection information and demand change information. The feature selection information includes the factor feature selection coefficient. The demand change information includes the processing efficiency floating value. After collection, the factor feature selection coefficient and the processing efficiency floating value are calibrated as YST and SCX respectively. Obtain the enterprise operation risk monitoring related data within a unit time, obtain the number of changes CS of the operation prediction model, obtain the change data set of the operation prediction model, obtain the number of risk categories TS of the change data set, and calculate the purity of the feature data. The calculation formula is: c is the number of categories, p r Indicates the proportion of the data of the rth category in the transition data set to the total data set. Perform purity analysis on the transition nodes of each transition number, establish a feature purity set GDu = {GD1, GD2, ..., GDw}, w is a positive integer, and calculate the feature purity mean Calculate the factor feature selection coefficient, the calculation expression is: Obtain the business data volume for relevant historical risk business analysis and establish a business data volume set, obtain the time required for business analysis and establish a business analysis time set, take the quotient of business data volume and business analysis time as business efficiency, establish a business efficiency set XLk = {XL1, XL2, ..., XLd}, d is a positive integer, and obtain the mean of the business efficiency set Get the real-time business processing efficiency SXL of the model state transition. The calculation formula of the processing efficiency floating value is:

4. The enterprise operation early warning monitoring method based on data analysis according to claim 3 is characterized in that: The obtained factor feature selection coefficient and processing efficiency floating value are dimensionless, and the risk warning index is established after removing the unit, and calibrated as F z , based on the formula: In the formula, e1 and e2 are the factor feature selection coefficients LB d And the area purchase floating value GM f The preset proportional coefficient, and e1 and e1 are both greater than 0.

5. The enterprise operation early warning monitoring method based on data analysis according to claim 4 is characterized in that: Compare the risk warning index with the data risk threshold to generate risk data signals and standard recommendation signals. The specific steps are as follows: Compare the risk warning index with the data risk threshold; If the risk warning index is less than the data risk threshold, a standard data signal is generated for the analyzed business data; If the risk warning index is greater than or equal to the data risk threshold, a risk data signal is generated for the analyzed business data.

6. The enterprise operation early warning monitoring method based on data analysis according to claim 5 is characterized in that: Combined with the status of the operational forecast model, joint warning is carried out. The specific steps are as follows: Confirm the business data that generates risk data signals and analyze the relationship between the operational forecast model status and business data; When the state of the operation prediction model changes from abnormal state and fault state to normal state, re-analyze the business data and contact relevant maintenance personnel to process the business data; When the state of the operational prediction model changes from a normal state to an abnormal state and a fault state, a corresponding category warning is issued for the category of the business data that generates the risk data signal, and the enterprise is notified to adjust its strategy; When the business data generates a normal data signal and the state of the operation prediction model changes from a normal state to an abnormal state and a fault state, a model warning is issued to the operation prediction model.

7. An enterprise operation early warning monitoring system based on data analysis, used to implement the method described in any one of claims 1 to 6, characterized in that: Including data acquisition module, data processing module and data early warning module; The data collection module is used to obtain enterprise data to establish an operation prediction model, collect operation stability information in the operation prediction model, analyze the status of the operation prediction model, classify and mark the operation prediction model into different states according to the analysis results, and send the collected data to the data processing module; The data processing module receives the data sent by the data acquisition module, analyzes the state transition of the classification and labeling operation prediction model, generates a risk warning index based on the transition details of the business data under the transition state, and sends it to the data warning module; The data warning module receives the data sent by the data processing module, compares the risk warning index with the data risk threshold, generates risk data signals and standard recommendation signals, and conducts joint warnings based on the status of the operation prediction model.

Images