Self-adaptive risk identification and data integration system based on statistical modeling

Abstract

The invention relates to the field of data analysis in enterprise risk management, in particular to an adaptive risk target set construction and data integration system based on statistical modeling. The system comprises a data aggregation module, a data asset management module, a business target management module, a business index management module, a business index distribution management module, a business model management module, a business target distribution module, a risk preference setting module, a risk matrix module and a risk management database table module. The system quantifies the uncertainty of business targets and indexes, identifies business risks and key risk factors, generates a risk matrix to display the possibility and influence degree of the business risks, and finally generates a database table structure for risk management, thereby improving the efficiency and accuracy of enterprise risk management.

Description

Technical Field

[0001] The present invention relates to the field of data analysis for risk identification in enterprise risk management. Specifically, it relates to an adaptive database construction system based on statistical modeling. Background Art

[0002] Enterprise Risk Management (ERM) has now become a recognized good management practice. With the development and wide application of information technology, the construction of information systems for key business operations in enterprises has matured, laying a solid foundation for the digital transformation of risk management. The primary task of risk management is risk identification. To identify risks using existing data, it is necessary to deeply explore the logic hidden behind the data. By establishing statistical models of business operations to quantify the magnitude and distribution law of the uncertainty of business objectives, enterprises can effectively utilize data assets to identify business risks and relevant factors that have a greater impact on risks. Currently, the biggest challenge faced by the digital transformation of risk management is the severe lack of interdisciplinary professionals who possess data analysis capabilities and master enterprise risk management theory and practice. In addition, current information-based products for enterprise risk management are far from complete risk management in terms of quantitative analysis and functional systems, and it is difficult to meet actual needs.

[0003] In the process of the digital transformation of enterprise risk management, there are still many challenges and problems, which are mainly reflected in the following aspects:

[0004] 1) It is difficult to analyze existing business data using quantitative methods for risk identification. Currently, many enterprises mainly rely on the experience and intuition of managers to identify potential risk areas and then include these areas in the scope of risk management. This experience-based method lacks scientificity and systematicness, resulting in insufficient accuracy and comprehensiveness of risk identification. For example, when a certain business operation has multiple key business objectives, it is difficult to subjectively evaluate the distribution of the uncertainty of all key objectives, and thus it is difficult to quantitatively compare the results of risk identification. When the risk management cost is limited, it is difficult for decision-makers to make a choice.

[0005] 2) It is difficult to analyze existing business data using quantitative methods for analyzing key factors affecting risks. Enterprise business systems usually contain rich database tables and business fields, but currently mainly rely on expert experience to select fields considered key as risk factors. This method cannot quantitatively evaluate the specific impact degree of risk factors on key business objectives. For example, a certain business system may contain a large number of fields such as transactions, warehousing, unit prices, and responsible departments. Risk management personnel cannot identify the impact degree of these fields on key business objectives, and thus it is difficult to determine their actual impact on key business objectives.

[0006] The problems of the existing technology in the digital transformation of risk management can be summarized as follows:

[0007] 1) Lack of systematic and standardized risk identification methods. Risk identification methods relying on experience are difficult to comprehensively cover all potential risks, and are highly subjective and easily affected by personal judgment.

[0008] 2) Lack of scientific basis for the selection of risk factors. The current risk factor analysis mainly relies on expert experience, lacks quantitative evaluation means, and is difficult to accurately judge the impact of risk factors on the business and its uncertainty.

[0009] 3) Insufficient data integration and analysis capabilities. Existing systems are difficult to effectively integrate and analyze scattered business data, and systematically process the data related to business objectives and their influencing factors from the perspective of risk management, resulting in lack of systematicness, accuracy and low efficiency in risk management.

[0010] Solution: To solve the above problems, the present invention provides a risk identification system based on statistical methods, which can systematically and quantitatively analyze business data by establishing a business model, so as to identify business risks with greater uncertainty and their key risk factors, automatically generate corresponding database table structures, and collect and process scattered business data from the perspective of risk management. Summary of the Invention

[0011] To solve the problems of risk identification and risk factor analysis existing in the prior art, the present invention provides an adaptive risk identification and data integration system based on statistical methods. By establishing a business model, it can systematically and quantitatively analyze business data, so as to identify business risks with greater uncertainty and their key risk factors. The system automatically forms a business risk management library table composed of business objectives and key risk factors required for risk management by aggregating various business data. The system includes the following main modules: a data management module, a model construction module, and a risk identification module.

[0012] According to one aspect of the present invention, there is provided an adaptive risk identification and data integration method based on statistical methods, as Figure 1 shown, characterized in that it includes:

[0013] 1) Obtain the existing business data of the enterprise. The acquisition objects of the existing data include data from existing business systems, offline data such as Excel and CSV, and data that can be provided by other external interfaces. The forms of the existing data include master data, business data, transaction data, analysis data, metadata, reference data, etc.

[0014] 2) Clean, transform, and manage the converged data for data asset management. The data cleaning processes the noise and missing values in the data to ensure data quality; data transformation converts the data into the format required by the system to ensure data consistency and availability; data governance maintains data consistency and integrity to ensure data reliability throughout the analysis process. Data asset management is for facilitating the reference to data when constructing business models and for clarifying the data sources of risk management library tables.

[0015] 3) Identify and define the business objectives of the enterprise, which is the basis for establishing statistical models. The identification and definition of the business objectives of the enterprise clarify certain business management requirements or strategic objectives targeted by risk identification. The business management requirements or strategic objectives can be qualitatively defined in text form, and business data can be extracted through data collection rules. After determining the business objectives, it provides a clear direction and goal for model construction. The business objectives are the final output variables of the model, and their statistical characteristics are used to quantitatively describe the distribution of business risks.

[0016] 4) Identify and map the key business indicators related to the business objectives, which are the basis for quantitative analysis. The key business indicators are the input variables of the quantitative indicators around the business objectives, and these variables are the data obtained on the basis of acquiring business data and effectively cleaning, transforming, and managing it. By quantifying business indicators, the uncertainty and risks of business objectives can be better understood.

[0017] 5) Quantify the distribution law of business indicators and conduct systematic analysis of business indicators through statistical methods. The quantification of the distribution law of business indicators records the analysis results such as the distribution type, maximum value, minimum value, mean, standard deviation, probability density, and sampling simulation data of business indicators. The quantitative management of business indicators helps identify the uncertain factors affecting business objectives.

[0018] 6) Establish a business model based on business objectives and business indicators. The business model refers to the logical relationship of the business from input variables to output variables.

[0019] 7) Analyze the distribution of business objectives according to the established business model. Through the logical relationship, the uncertainty of business indicators can be effectively transmitted to business objectives, thereby quantifying the uncertainty and its distribution law of business objectives and the impact degree of business indicators on business objectives, providing a scientific basis for risk identification.

[0020] 8) The system allows users to set risk preferences to assist in identifying and evaluating business risks. The setting of risk preferences means that users set the business objectives, i.e., the confidence level and value range of the output variables, according to their own acceptable degree of business uncertainty, and specify the unacceptable confidence level or value range. The setting of the confidence level and value range is interrelated. Users can only set one of them, and the other will be automatically adjusted according to the setting.

[0021] 9) Generate a risk matrix based on the analysis results to display the likelihood and impact degree of business risks. The risk matrix is a risk management tool that evaluates and displays the likelihood of risks and the severity of their consequences through a two-dimensional table. The likelihood refers to the probability density corresponding to the unacceptable confidence level or value range of the business objectives specified by the user. If this probability density exceeds 70%, it is marked as "high likelihood"; if the probability density is between 50% - 70%, it is "medium likelihood"; if the probability density is less than 50%, it is "low likelihood". The impact degree is set to three levels of "high, medium, low" according to expert opinions. The risk matrix provides an overall risk assessment perspective for enterprises, facilitating risk management decisions.

[0022] 10) The system finally automatically generates a database table structure for risk management according to the above process. It includes a business objective table, a business indicator table, and a risk management object table. The business objective table is a table structure composed of data extraction rules defined for business objectives. Through this table structure, users can penetrate business data. The business indicator table is the data fields defined for key business indicators. Through this table structure, users can penetrate business data. The risk management object table is a table structure obtained by cross - querying the business objective table and the business indicator table. Through this table structure, users can penetrate business data. Users can also publish the above data in the form of an interface for further data utilization.

[0023] According to one aspect of the present invention, there is provided an adaptive risk identification and data integration system based on statistical methods, as Figure 1 shown, characterized by comprising:

[0024] 1) A data aggregation module for obtaining the existing business data of the enterprise, and the data includes internal system data, Excel files, CSV files, and data obtained through other external interfaces.

[0025] 2) A data asset management module for cleaning, transforming, and governing the aggregated data. The data cleaning is to process the noise and missing values in the data; the data transformation is to convert the data into the format required by the system; the data governance is to maintain the consistency and integrity of the data.

[0026] 3) Business objective management module, which is used to identify and define the business objectives of an enterprise. The business objectives can qualitatively define the business management requirements or strategic objectives in text form, and extract business data through data collection rules.

[0027] 4) Business indicator management module, which is used to identify and map the key business indicators related to business objectives. These indicators are the basis for quantitative analysis.

[0028] 5) Business indicator distribution management module, which is used to systematically analyze business indicators through statistical methods and quantify the distribution law of business indicators. The analysis includes the distribution type, maximum value, minimum value, mean value, standard deviation, probability density and sampling simulation data of business indicators, etc.

[0029] 6) Business model management module, which establishes a business model based on business objectives and business indicators. The business model refers to the logical relationship of a business from input variables to output variables. The establishment of a business model includes steps such as variable selection, data preprocessing, model selection, model training, model verification, model optimization, model deployment and model maintenance, etc.

[0030] 7) Business objective distribution module, which is used to analyze the distribution of business objectives according to the established business model. By transmitting the uncertainty of business indicators to business objectives, the uncertainty of business objectives and its distribution law are quantified.

[0031] 8) Risk preference setting module, which allows users to set the confidence level or value range of business objectives (i.e., output variables) according to their acceptable degree of business uncertainty. The confidence level and value range settings are interrelated. Users can only set one of them, and the other will be automatically adjusted.

[0032] 9) Risk matrix, which shows the possibility and impact degree of business risks according to the analysis results. The risk matrix evaluates and shows the possibility of risks and the severity of their consequences through a two-dimensional table. The possibility is determined by the probability density corresponding to the unacceptable confidence level or value range of business objectives, and the impact degree is set to three levels of "high, medium, low" according to expert opinions.

[0033] 10) Risk management library table module, which is used to generate a business risk objective set, a business risk factor set, and publish the identification results.

[0034] Through the collaborative work of the above functional modules, the system of the present invention can effectively solve the deficiencies of existing risk management methods and improve the efficiency and accuracy of enterprise risk management. The system can not only automatically generate the database table structure, collect and process scattered business data, but also provide scientific risk identification and management tools for enterprises to help enterprises achieve the digital transformation of risk management. Description of the Drawings

[0035] Visualization of the business model The accompanying drawings described herein are used to provide a further understanding of the present invention and form a part of this application. The illustrative embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the accompanying drawings:

[0036] Figure 1 is a schematic diagram of an adaptive database construction system based on statistical modeling of the present invention;

[0037] Figure 2 is a schematic diagram of constructing an end-of-period inventory quantity model in Embodiment 1 of the present invention;

[0038] Figure 3 is a schematic diagram of the average inventory risk identification result in Embodiment 1 of the present invention; Detailed implementation manners

[0039] The present invention will be described in detail below with reference to the accompanying drawings and in combination with two embodiments. It should be noted that, without conflict, the embodiments in this application and the features in the embodiments may be combined with each other.

[0040] According to Embodiment 1 of the present invention, a risk identification application for an enterprise's order quantity is provided, as Figure 1 shown. In the embodiment of the present invention, the data aggregation module 01 acquires a large amount of original warehouse data, including the inventory, inbound and outbound, procurement requirements, etc. of various materials, and cleans, transforms, and manages these original data through the data asset management module 02. The processed data is further analyzed by the business objective management module 03, the business indicator management module 04, and the business indicator distribution management module 05 to generate standardized data, key business indicators, and the distribution rules of business indicators. Subsequently, the business model management module 06 and the business objective distribution module 07 establish corresponding business models and obtain the distribution of business objective uncertainty based on the above-mentioned multiple standardized data, key business indicators, and the distribution rules of business indicators. The risk preference setting module 08 is used to set according to the preference for the business objective risk, and finally generates the risk matrix of the business objective and the database table structure around the business risk management objective.

[0041] The specific steps of this example are as follows:

[0042] Step 1: The data aggregation module 01 acquires the original data on the material procurement business in the enterprise's material management information system. The original data of the procurement business is mainly data such as the inventory, inbound and outbound, and procurement requirements of various materials. For the enterprise's internal informatization ecosystem, the data aggregation module 01 supports direct connection to the business system database, API communication, and offline data aggregation such as EXCEL and CSV. In this example, the data acquisition module 01 acquires data such as inventory and inbound and outbound of the material procurement business.

[0043] Step 2: Based on the raw data obtained by the data aggregation module 01, the data asset management module 02 removes duplicate information, error information, and useless information from the element data to obtain standard variables.

[0044] Original data Include fields Inbound and outbound Inbound and outbound material categories, inbound and outbound time, inbound and outbound quantity Inventory Real-time inventory quantity

[0045] Step 3: The business objective management module 03 clarifies the core management objectives of the material procurement business. The business objectives can qualitatively define the business management requirements or strategic objectives in text form. In this example, the core management objective of the material procurement business is defined as: "Controlling the average inventory level within a certain limit considering the order - receipt time, inventory inspection interval days, and service level conditions."

[0046] Step 4: The business indicator management module 04 analyzes the standard variables and / or the relationships between multiple standard variables, and generates processed variables according to specific logical rules required by the business model. The logical rules include but are not limited to basic variable processing, composite variable processing, SQL variable processing, parameter configuration, etc. In this example, the processed variables obtained through the data asset module 02 based on the raw data are shown in the following table:

[0047] Original variables Logical rules Processed variables Inbound and outbound variables Based on the original data, obtain the inventory quantity on Monday of each week Beginning inventory Inventory variables, inbound and outbound variables Based on the original database inventory variables and inbound and outbound variables, calculate the inventory at the end of each Sunday Ending inventory Demand, beginning inventory Subtract the beginning inventory from the demand, and take the minimum value as 0 Shortage quantity None Through parameter configuration, set as a random variable subject to a certain distribution Demand None Through parameter configuration, set as a fixed value, unit: week Order - receipt time None Through parameter configuration, set as a fixed value, unit: week Inventory check interval days None Through parameter configuration, set as a fixed value in (0, 1), (0.98) Service level per period Demand per period and its standard deviation, order - receipt time, inventory check interval days, service level per period Fixed - order - quantity method Order quantity

[0048] Step 5: The business indicator distribution management module 05 further processes the obtained business indicators and quantifies their uncertainties. In this example, key variables such as the demand per period, the standard deviation of the demand per period, inventory inspection interval days, order - receipt time, service level per period, beginning inventory, ending inventory, shortage quantity, and order quantity are defined. Except for the demand per period which is an uncertain input variable, the other items are either processed from the raw data or are fixed values. Through the business indicator management module 04, the distribution and distribution parameters that the demand per period follows can be defined, and it is defined that the demand per period ~ N(100, 20). In this example, the business indicator set is shown in the following table:

[0049] Variable name Data sample Data type Demand per period 100 Integer type Standard deviation of demand per period 20 Integer type Inventory check interval days 2 Integer type Order - receipt time 1 Integer type Service level per period 0.98 Decimal type Beginning inventory 500 Integer type Ending inventory 400 Integer type Shortage quantity 0 Integer type Order quantity 0 Integer type

[0050] Step 6: The business model management module 06 establishes a business model based on the business indicator variable set and identifies the uncertainties of the business objectives. In this example, the business model management module 06 constructs the logical relationships between all business indicators and business objectives within each calculation period, such as Figure 2 shown that the average inventory level can be obtained from the ending inventory of the nth period.

[0051] Step 7: The business objective distribution module 07 transmits the uncertainty of the business metrics to the business objectives according to the business model, quantifies the uncertainty of the business objectives and its distribution law, and the business objective distribution module 07 records the uncertainty of the business objectives and its distribution law. In this example, the business objective is "to control the average inventory level within a certain limit". Therefore, the detailed statistics of the calculated business objectives to be recorded include: minimum value, maximum value, average value, standard deviation, variance, skewness, kurtosis, mode, 5% - 95% statistics, as shown in the following table:

[0052] Name Average inventory Minimum value 408.5794 Maximum value 1639.698 Average value 749.8268 Standard deviation 151.8936 Variance 23071.66 Skewness 0.8866303 Kurtosis 4.45135 Mode 676.5338 5% (Percentage) 539.6302 10% (Percentage) 578.3827 15% (Percentage) 602.6314 20% (Percentage) 621.6215 25% (Percentage) 641.0514 30% (Percentage) 659.8323 35% (Percentage) 675.0256 40% (Percentage) 693.2773 45% (Percentage) 711.3625 50% (Percentage) 729.6042 55% (Percentage) 746.8147 60% (Percentage) 766.8904 65% (Percentage) 788.7175 70% (Percentage) 810.0966 75% (Percentage) 836.9243 80% (Percentage) 866.5646 85% (Percentage) 903.6638 90% (Percentage) 948.3779 95% (Percentage) 1026.849

[0053] Step 8: The risk preference setting module 08 sets the confidence level or value range of the business objective (i.e., the output variable) according to the acceptable degree of the uncertainty of the business objective. In this example, the "average inventory limit" is set to 730, and the business objective is "to control the average inventory level within 730". It can be understood that the risk preference regarding the business objective is: it is unacceptable when the average inventory level is greater than 730. From the detailed statistics of the business objective obtained by the objective distribution module 07, it can be seen that the probability of the business objective triggering the risk preference reaches 50%, that is, the probability that the average inventory level is greater than 730 is greater than or equal to 50%.

[0054] Step 9: According to the system - preset risk possibility threshold range, a probability density between 50% - 70% is "medium possibility". Combining with the expert's assessment of the impact degree on the business objective as "medium" impact degree. The identification result of the risk regarding the business objective is medium - risk. The risk matrix is as Figure 3 shown.

[0055] Step 10: The risk management library table module 10 automatically generates the database table structures of the risk objectives and business risk factors according to the above - mentioned process. These tables can be published as APIs for the internal systems of the enterprise to apply the risk identification results. In this example, the generated database tables are as follows:

[0056] Risk Goals Table (RiskGoals)

[0057] Serial number Field name Type Width Key Description 1 GoalID INT Primary key Unique identifier of the risk goal 2 GoalName VARCHAR 255 Name of the risk goal 3 GoalDescription TEXT Description of the risk goal 4 GoalType VARCHAR 100 Type of the risk goal 5 GoalValue FLOAT Value of the risk goal 6 ConfidenceLevel FLOAT Confidence level of the risk goal 7 AcceptableRangeStart FLOAT Starting value of the acceptable range of the risk goal 8 AcceptableRangeEnd FLOAT Ending value of the acceptable range of the risk goal 9 CreatedDate TIMESTAMP Creation date, default value is the current time 10 ModifiedDate TIMESTAMP Update date, default value is the current time, automatically updated when updated

[0058] Business Risk Indicators Table (RiskIndicators)

[0059] Serial number Field name Type Width Key Description 1 IndicatorID INT Primary key Unique identifier of the business risk factor 2 IndicatorName VARCHAR 255 Name of the business risk factor 3 IndicatorDescription TEXT Description of Business Risk Factors 4 IndicatorType VARCHAR 100 Type of Business Risk Factor 5 GoalID INT Foreign Key Associated Risk Goal ID, referring to RiskGoals(GoalID) 6 DataType VARCHAR 50 Data Type 7 DataValue FLOAT Data Value 8 MinValue FLOAT Minimum Value of Data 9 MaxValue FLOAT Maximum Value of Data 10 MeanValue FLOAT Average Value of Data 11 StandardDeviation FLOAT Standard Deviation of Data 12 ConfidenceLevel FLOAT Confidence Level 13 CreatedDate TIMESTAMP Creation Date, default value is the current time 14 ModifiedDate TIMESTAMP Update Date, default value is the current time, automatically updated when updated

Claims

1. An adaptive risk identification and data integration method based on statistical modeling, characterized in that: The following steps are involved: Acquire the company's existing business data; clean, transform and govern the aggregated data; identify and define the company's business goals; identify and map key business indicators related to business goals; quantify the distribution patterns of business indicators; Establish business models and analyze the distribution of business objectives; set risk preferences, identify and assess business risks; generate risk matrices based on analysis results to demonstrate the likelihood and impact of business risks; Generate a database table structure for risk management, including a business objective table, a business indicator table, and a risk management object table.

2. The method according to claim 1, characterized in that The data acquisition objects include existing business system data, offline data such as Excel and CSV, and data provided by other external interfaces.

3. The method according to claim 1, characterized in that The data cleaning is to deal with noise and missing values ​​in the data to ensure data quality; data conversion is to convert data into the format required by the system to ensure data consistency and availability; data governance is to maintain data consistency and integrity to ensure data reliability throughout the entire analysis process.

4. The method according to claim 1, characterized in that: The possibility of the risk matrix refers to the probability density corresponding to the unacceptable confidence level or value range of the business objective, and the degree of impact is set at three levels of "high, medium, and low" based on expert opinions.

5. An adaptive risk identification and data integration system based on statistical modeling, characterized in that: It includes the following modules: a data aggregation module, which is used to obtain the existing business data of the enterprise, including internal system data, Excel files, CSV files and data obtained through other external interfaces; The data asset management module is used to clean, transform and govern the aggregated data. Data cleaning is to process the noise and missing values ​​in the data, data transformation is to convert the data into the format required by the system, and data governance is to maintain the consistency and integrity of the data. The business goal management module is used to identify and define the business goals of the enterprise. Business goals can be used to qualitatively define business management requirements or strategic goals in text form, and extract business data through data collection rules; the business indicator management module is used to identify and map key business indicators related to business goals. These indicators are the basis for quantitative analysis; the business indicator distribution management module is used to systematically analyze business indicators through statistical methods and quantify the distribution rules of business indicators; The business model management module is used to establish a business model based on business objectives and business indicators. The business objective distribution module is used to analyze the distribution of business objectives based on the established business model. The risk preference setting module allows users to set the confidence level or value range of business objectives based on their own acceptance of business uncertainty; The risk matrix module is used to generate a risk matrix based on the analysis results to show the possibility and impact of business risks; The risk management database table module is used to generate business risk target sets, business risk factor sets, and publish identification results.

6. The system according to claim 5, characterized in that The data aggregation module supports direct connection to the business system database, API communication, and offline data aggregation such as Excel and CSV.

7. The system according to claim 5, characterized in that The data asset management module includes three sub-modules: data cleaning, data conversion and data governance.

8. The system according to claim 5, characterized in that The business objective management module extracts business data through data collection rules to clarify the business management requirements or strategic objectives for risk identification.

9. The system according to claim 5, characterized in that The business indicator management module is used to quantify business indicators and analyze the distribution patterns of business indicators through statistical methods.

10. The system according to claim 5, characterized in that The business model management module includes steps such as variable selection, data preprocessing, model selection, model training, model verification, model optimization, model deployment and model maintenance.

11. The system according to claim 5, characterized in that The risk preference setting module allows users to set the confidence level or value range of business objectives to identify and assess business risks.

12. The system according to claim 5, characterized in that The risk matrix evaluates and displays the likelihood of risks and the severity of their consequences in a two-dimensional table.

Images