Device and method for predicting default probability of each company on basis of employment index

Abstract

A device for predicting the default probability of each company on basis of an employment index comprises: a data collection unit for collecting employment index data indicating an employment state of each company; a feature variable extraction unit for extracting, from the employment index data, a plurality of feature variables for predicting default probability; a model construction unit for learning the plurality of feature variables to pursue a prediction model for predicting the default probability; and a default probability prediction unit for predicting the default probability of a specific company by using the prediction model.

Description

Device and method for predicting corporate default probability based on employment indicators

[0001] The present invention relates to a device for predicting the probability of bankruptcy for each company, and more specifically, to a device and method for predicting the probability of bankruptcy for each company based on an employment index, which can predict the probability of bankruptcy for a specific company by collecting employment index data indicating the employment status for each company and extracting feature variables for predicting the probability of bankruptcy.

[0002]

[0003] Typically, the method of predicting default is to utilize financial information in a company's financial statements, and this financial information is used to indirectly predict default by measuring the company's credit rating.

[0004] These conventional techniques use characteristic values ​​such as liabilities-retained earnings / tangible assets based on financial statements as indicators for evaluating the stability of a company, but these formulas are generally determined by experts, making it difficult to determine their validity. In addition, in order to realistically increase the accuracy of bankruptcy prediction, it is necessary to consider the influence of external factors as well as the company's internal information, but conventional techniques have the problem of not considering the influence of these external factors.

[0005] Therefore, a method is required to predict corporate bankruptcy by comprehensively analyzing such factors.

[0006]

[0007] [Prior Art Literature]

[0008] [Patent Document]

[0009] Korean Patent No. 10-2018-0037358 (March 30, 2018)

[0010]

[0011] One embodiment of the present invention provides a device and method for predicting the probability of bankruptcy for each company based on employment indices, which can collect employment indices data for each company and extract characteristic variables for predicting the probability of bankruptcy.

[0012] One embodiment of the present invention provides a device and method for predicting corporate default probability based on employment indicators, which can learn a plurality of feature variables and pursue a prediction model for predicting default probability.

[0013] One embodiment of the present invention provides a device and method for predicting corporate default probability based on employment indicators, which can construct a machine learning model that generates as output a continuous default probability of a specific corporate at each future point in time.

[0014] One embodiment of the present invention provides a device and method for predicting the probability of bankruptcy of a specific company based on employment indicators.

[0015]

[0016] According to one embodiment of the present invention, a device for predicting corporate default probability based on employment index includes: a data collection unit for collecting employment index data indicating employment status for each company; a feature variable extraction unit for extracting a plurality of feature variables for predicting default probability from the employment index data; a model construction unit for learning the plurality of feature variables to pursue a prediction model for predicting default probability; and a default probability prediction unit for predicting the default probability of a specific company using the prediction model.

[0017] The above data collection unit can collect the employment index data continuously aggregated for at least a specific period of time for each company.

[0018] The above-mentioned feature variable extraction unit can extract the number of national pension data for each company in the current month, the number of new employees in the current month, the number of resignations in the current month, the number of employees in the current month, the amount of notice in the current month, the average salary, the employment volatility, MoM, YoY, and the Z-score by mid-class industry as the above-mentioned multiple feature variables.

[0019] The above model construction unit can construct a machine learning model as the prediction model by receiving the plurality of feature variables as input and generating the continuous probability of default of the company at each future point in time as output.

[0020] The above model building unit removes outliers for each feature variable of the learning data, sets a rating criterion for each feature variable according to the distribution of the learning data, classifies the verification data by applying the rating criterion to the verification data, assigns a partial score within a specific range according to the number of ratings for each feature variable, and calculates the final score by adding up the rating scores of each of the plurality of feature variables.

[0021] The above model building unit can classify the grade of the verification data so that the higher the grade of the default rate, the higher the grade of the learning data.

[0022] The above default probability prediction unit can calculate a default variable grade and a final score based on the default probability output by the above prediction model.

[0023] In one embodiment, a method for predicting corporate default probability based on an employment index may include a step of collecting employment index data indicating the employment status of each company through a data collection unit; a step of extracting a plurality of feature variables for predicting default probability from the employment index data through a feature variable extraction unit; a step of learning the plurality of feature variables through a model construction unit to pursue a predictive model for predicting the default probability; and a step of predicting the default probability of a specific company using the predictive model through a default probability prediction unit.

[0024]

[0025] The disclosed technology may have the following effects. However, this does not mean that a particular embodiment must include all or only the following effects, and thus the scope of the disclosed technology should not be construed as being limited thereby.

[0026] A device and method for predicting corporate default probability based on employment indicators according to one embodiment of the present invention can collect employment indicator data for each company and extract feature variables for predicting corporate default probability.

[0027] A device and method for predicting corporate default probability based on employment indicators according to one embodiment of the present invention can learn a plurality of characteristic variables and pursue a prediction model for predicting default probability.

[0028] A device and method for predicting corporate default probability based on employment indicators according to one embodiment of the present invention can construct a machine learning model that generates as output a continuous default probability for a specific corporate entity at each future point in time.

[0029] A device and method for predicting the probability of bankruptcy of a specific company based on an employment index according to one embodiment of the present invention can predict the probability of bankruptcy of a specific company.

[0030]

[0031] FIG. 1 is a diagram illustrating a system for predicting the probability of bankruptcy by company according to one embodiment of the present invention.

[0032] Figure 2 is a diagram explaining the configuration of the corporate default probability prediction system in Figure 1.

[0033] Figure 3 is a drawing explaining the configuration of the corporate default probability prediction device of Figure 1.

[0034] Figure 4 is a flowchart explaining the functional configuration of the corporate default probability prediction device of Figure 1.

[0035] FIG. 5 is a drawing illustrating one embodiment of a corporate default probability prediction device according to the present invention.

[0036]

[0037] The description of the present invention is merely an example for structural and functional explanation, and therefore, the scope of the present invention should not be construed as being limited by the embodiments described in the text. That is, since the embodiments can be modified in various ways and can take various forms, the scope of the present invention should be understood to include equivalents that can realize the technical idea. In addition, the purposes or effects presented in the present invention do not mean that a specific embodiment must include all of them or only such effects, and therefore, the scope of the present invention should not be construed as being limited thereby.

[0038] Meanwhile, the meaning of the terms described in this application should be understood as follows.

[0039] Terms such as "first" and "second" are intended to distinguish one component from another, and the scope of the rights should not be limited by these terms. For example, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component.

[0040] When a component is said to be "connected" to another component, it should be understood that while it may be directly connected to that other component, there may also be other components intervening. Conversely, when a component is said to be "directly connected" to another component, it should be understood that there are no other intervening components. Similarly, other expressions describing relationships between components, such as "between" and "directly between," or "adjacent to" and "directly adjacent to," should be interpreted similarly.

[0041] Singular expressions should be understood to include plural expressions unless the context clearly indicates otherwise, and terms such as "comprises" or "have" should be understood to specify the presence of a feature, number, step, operation, component, part or combination thereof, but not to exclude the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0042] For each step, the identifiers (e.g., a, b, c, etc.) are used for convenience of explanation and do not describe the order of the steps. The steps may occur in a different order than stated unless the context clearly dictates a specific order. That is, the steps may occur in the same order as stated, may be performed substantially simultaneously, or may be performed in the opposite order.

[0043] The present invention can be implemented as computer-readable code on a computer-readable recording medium. The computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system. Examples of the computer-readable recording medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices. Furthermore, the computer-readable recording medium can be distributed across network-connected computer systems, so that the computer-readable code can be stored and executed in a distributed manner.

[0044] Unless otherwise defined, all terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which this invention pertains. Terms defined in commonly used dictionaries should be interpreted to be consistent with their meaning within the context of the relevant technology, and should not be interpreted as having an idealized or overly formal meaning unless explicitly defined herein.

[0045]

[0046] FIG. 1 is a diagram illustrating a system for predicting the probability of bankruptcy by company according to one embodiment of the present invention.

[0047] Referring to FIG. 1, a corporate default probability prediction system (100) may include a user terminal (110), a corporate default probability prediction device (130), and a database (150).

[0048] A user terminal (110) may correspond to a terminal device operated by a user. According to one embodiment of the present invention, one or more users may correspond to one or more, and in the case of multiple users, they may be designated and distinguished into one or more user groups. Each of one or more users may correspond to one or more user terminals (110), and for example, a user may correspond to a person (i.e., a modeler) who requests data collection regarding employment indices and performs data modeling. Here, the employment index may correspond to an index including an employment status index, an unemployment index, an employment insurance new applicant index, an employment insurance insured person index, an employment insurance insured person increase / decrease rate in the number of employment insurance insured persons, etc. In addition, although FIG. 1 is expressed as one user terminal (110), a first user may correspond to a first user terminal, a second user may correspond to a second user terminal, …, an n-th user (where n is a natural number) may correspond to an n-th user terminal, respectively.

[0049] In one embodiment, the user terminal (110) may be a computing device operated by a user as a component of the corporate default probability prediction system (100). Here, the user may include a corporation or institution seeking to conduct an employment status analysis, or a corporation or institution conducting an analysis of corporate financial statements or credit risk-related data.

[0050] In one embodiment, the user terminal (110) may request corporate default probability data from the corporate default probability prediction device (130) or provide employment indicators and characteristic variables. Here, the characteristic variables may correspond to indices for predicting the probability of default based on the employment indicators. Further details regarding the characteristic variables are described below in FIG. 3. Furthermore, the user terminal (110) may install and execute a dedicated program or application to interface with the corporate default probability prediction device (130).

[0051] The corporate default probability prediction device (130) may be implemented as a server corresponding to a computer or program that performs the corporate default probability prediction service based on employment indicators according to the present invention. Furthermore, the corporate default probability prediction device (130) may be connected to a user terminal (110) via a wired network or a wireless network such as Bluetooth, WiFi, or LTE, and may transmit and receive data with the user terminal (110) via the network.

[0052] Additionally, the corporate default probability prediction device (130) can provide a corporate default probability prediction service based on employment indicators according to the present invention to a user terminal (110) through its own platform. For example, the corporate default probability prediction device (130) can provide a default probability prediction platform that provides visualized modeling for employment status analysis, and for this purpose, can be implemented to operate in connection with an independent external system (not shown in FIG. 1).

[0053] The database (150) may correspond to a storage device that stores various information required during the operation of the corporate default probability prediction device (130). For example, the database (150) may store documents containing information collected or processed in various forms from a user terminal (110) or a website such as Google during the process of performing corporate default probability prediction.

[0054] In addition, in FIG. 1, the database (150) is depicted as an independent device from the corporate default probability prediction device (130), but is not necessarily limited thereto, and may be implemented as a logical storage device included in the corporate default probability prediction device (130) based on employment indicators.

[0055]

[0056] Figure 2 is a diagram explaining the configuration of the corporate default probability prediction system in Figure 1.

[0057] Referring to FIG. 2, the corporate default probability prediction device (130) may include a processor (210), memory (230), user input / output unit (250), and network input / output unit (270).

[0058] The processor (210) can execute a service procedure for predicting the probability of corporate default based on employment indicators according to an embodiment of the present invention, manage the memory (230) that is read or written in the process, and schedule a synchronization time between volatile memory and non-volatile memory in the memory (230). The processor (210) can control the overall operation of the device for predicting the probability of corporate default (130), and is electrically connected to the memory (230), the user input / output unit (250), and the network input / output unit (270) to control the data flow therebetween. The processor (210) can be implemented as a CPU (Central Processing Unit) of the device for predicting the probability of corporate default (130).

[0059] The memory (230) may include an auxiliary memory device implemented with a non-volatile memory such as an SSD (Solid State Disk) or an HDD (Hard Disk Drive) and used to store all data required for the corporate default probability prediction device (130), and may include a main memory device implemented with a volatile memory such as a RAM (Random Access Memory). In addition, the memory (230) may store a set of commands for executing the corporate default probability prediction service based on employment indicators according to the present invention by being executed by an electrically connected processor (210).

[0060] The user input / output unit (250) includes an environment for receiving user input and an environment for outputting specific information to the user, and may include, for example, an input device including an adapter such as a touchpad, a touchscreen, a virtual keyboard, or a pointing device, and an output device including an adapter such as a monitor or a touchscreen. In one embodiment, the user input / output unit (250) may correspond to a computing device connected via remote access, and in such a case, the corporate default probability prediction device (130) may be performed as an independent server.

[0061] The network input / output unit (270) provides a communication environment for connecting to a user terminal (110) via a network, and may include, for example, an adapter for communication such as a Local Area Network (LAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), and a Value Added Network (VAN). In addition, the network input / output unit (270) may be implemented to provide a short-range communication function such as WiFi or Bluetooth, or a wireless communication function of 4G or higher for wireless transmission of data.

[0062]

[0063] Figure 3 is a drawing explaining the configuration of the corporate default probability prediction device of Figure 1.

[0064] Referring to FIG. 3, the corporate default probability prediction device (130) may include a data collection unit (310), a feature variable extraction unit (330), a model construction unit (350), a default probability prediction unit (370), and a control unit (390).

[0065] Here, the corporate default probability prediction device (130) does not need to include all of the above functional components simultaneously. Depending on the specific embodiment, some of the above components may be omitted, or some or all of the above components may be selectively included. Furthermore, the corporate default probability prediction device (130) may be implemented as an independent module that selectively includes some of the above components, and the corporate default probability prediction platform method according to the present invention may be implemented through interoperability between the modules. The operation of each component will now be described in detail.

[0066]

[0067] The data collection unit (310) can collect employment index data indicating the employment status of each company. Here, the employment index data may correspond to one index for evaluating the stability of the company, and may correspond to data related to the number of people employed by the company, and may further include, for example, the number of new hires in the current month, the number of resignations in the current month, the total number of employees in the current month, and the total national pension notification amount in the current month. In one embodiment, the data collection unit (310) can receive employment index data through the user terminal (110), and is not necessarily limited thereto, and may collect employment index data by interacting with an external website (e.g., the National Pension Service) through the user terminal (110) and store the employment index data in the database (150).

[0068] In one embodiment, the data collection unit (310) may collect employment index data continuously aggregated for at least a specific period of time for each company. For example, the data collection unit (310) may collect employment index data for one or more companies on a monthly basis and compile the employment index data for 12 months to generate annual employment index data. Here, the data collection unit (310) may select employment index data for a specific month based on input from the user terminal (110) and collect employment index data for the previous 12 months and employment index data for the next 12 months based on the selected employment index data. In one embodiment, the data collection unit (310) may analyze the history of each collected company and collect employment index data according to the history. For example, the data collection unit (310) may analyze the year of establishment for a specific company and collect employment index data for the entire or a portion of the company's history.

[0069] In one embodiment, the data collection unit (310) may, during the process of collecting employment indicator data, check whether the employment indicator data is continuous data for at least 12 months. Here, the data collection unit (310) may perform a check on the employment indicator data, and if the employment indicator data does not meet a preset condition (e.g., continuous data for at least 12 months), a data purification process may be performed to filter the employment indicator data. In other words, the data collection unit (310) may filter out data that does not meet the condition among one or more employment indicator data, thereby improving the reliability of the employment indicator data.

[0070] The feature variable extraction unit (330) can extract multiple feature variables for predicting the probability of default from employment index data. Here, the feature variables may correspond to variables used for predicting default among the employment index data, and may include, for example, national pension data, the number of corporate employees, the number of new hires, and the number of resignations. The feature variable extraction unit (330) can extract at least two feature variables for predicting the probability of default from the employment index data and store the feature variables in the database (150). In addition, the feature variable extraction unit (330) can generate new feature variables by combining the feature variables stored in the database (150).

[0071] In one embodiment, the feature variable extraction unit (330) may extract, as multiple feature variables, the number of national pension data for each company, the number of new hires for the same month, the number of new hires for the same month, the number of employees for the same month, the amount of notice for the same month, the average salary, employment volatility, MoM, YoY, and the Z-score by mid-class industry. Here, the national pension data for the same month may correspond to data that compiles the national pension insurance premiums generated from the salaries of employees of a specific company monthly. The number of new hires for the same month may correspond to a monthly count of the number of employees hired by the company, and the number of new hires for the same month may correspond to a monthly count of the number of employees who have left the company. In addition, the number of employees for the same month may correspond to a monthly count of the number of employees working for the company, and the amount of notice for the same month may correspond to the monthly provision of expenses incurred by the company, such as salaries and monthly rent incurred on company premises, and may correspond to all expenses incurred in the process of operating the company. The average salary may correspond to the average salary paid to employees in the current month, but is not necessarily limited to this. It can also be calculated by calculating the total salary for the current month from the amount notified in the current month and then dividing it by the total salary for the current month / total number of employees. Furthermore, the employment volatility trend may correspond to data obtained by combining data on the number of new hires and new quits in the current month and dividing the data by the number of employees in the current month. It is not necessarily limited to this, but can also correspond to data generated by combining data on the number of new hires, new quits, and new employees in the current month. MoM may correspond to growth compared to the previous month, and YoY may correspond to growth compared to the same period in the previous year.In addition, Z-score can be a number that statistically creates a normal distribution and shows the position of each case on the standard deviation, and is not necessarily limited to this, and can correspond to data calculated by industry by standardizing the values ​​by utilizing the average and standard deviation for each of the national pension data for the current month, the number of new employees for the current month, the number of resignations for the current month, the number of employees for the current month, the amount of notice for the current month, the average salary, employment volatility, MoM, and YoY.

[0072] In one embodiment, the feature variable extraction unit (330) may define the Z-score as follows.

[0073] [Mathematical Formula 1]

[0074]

[0075]

[0076] Here, may correspond to the data values ​​of the feature variables, may correspond to the mean of the feature variable, and s may correspond to the standard deviation.

[0077]

[0078] In one embodiment, the feature variable extraction unit (330) may define the average salary for 12 months as follows.

[0079] [Equation 2]

[0080]

[0081]

[0082] Here, AMT may correspond to the current month's notice amount, and NBM_SBS may correspond to the current month's employees.

[0083] The model construction unit (350) can learn a plurality of feature variables to pursue a prediction model for predicting the probability of default. Here, the prediction model can be constructed by using the feature variables generated by the feature variable extraction unit (330) as learning data. That is, the model construction unit (350) can construct a corporate default probability prediction model by learning the feature variables generated through the feature variable extraction unit (330). Here, the model construction unit (350) can selectively apply a learning algorithm according to the feature variables, and can independently apply a plurality of learning algorithms as needed to construct a plurality of corporate default probability prediction models.

[0084] In one embodiment, the model building unit (350) may build a machine learning model as a prediction model by receiving a plurality of feature variables as input and generating a continuous default probability for each future point in time of the corresponding company as output. For example, the model building unit (350) may receive a default probability prediction period and at least one analysis target company from the user terminal (110) and perform an analysis based on the feature variables, thereby continuously generating default probabilities for the analysis target company after 1 month, after 2 months, ..., after 6 months. Here, the model building unit (350) may perform a monthly time series analysis by expressing the default probability for at least one company as a graph based on a plurality of feature variables. In one embodiment, the model building unit (350) may visually express the default probability for each company by differentially assigning colors according to the degree of default probability. For example, if the default probability is 80% or higher, the company may be displayed in red on the user terminal (110) to express the risk status.

[0085] In one embodiment, the model building unit (350) removes outliers for each feature variable of the training data, sets a ranking standard for each feature variable according to the distribution of the training data, classifies the testing data by applying the ranking standard to the testing data, assigns a partial score within a specific range according to the number of grades for each feature variable, and calculates a final score by adding up the ranking scores of each of the multiple feature variables. Here, the ranking standard may correspond to a standard for classifying each feature variable into 10 grades, but is not necessarily limited thereto and may correspond to a standard for classifying each feature variable according to the distribution of the training data. In one embodiment, the model building unit (350) may remove outliers in the training data based on a value having a Z-score of 3 in the process of removing outliers in the training data. That is, when the Z-score of the training data is higher or lower than 3, the model building unit (350) may determine whether the training data is abnormal and remove the outlier.

[0086] In one embodiment, the model building unit (350) may utilize at least a portion of the received training data as training data and determine the remainder as validation data. For example, the model building unit (350) may receive 12 months of data for each company, train the first three months of the data, and utilize the subsequent nine months of data as validation data. In one embodiment, the model building unit (350) may assign a score of 1 to 10 to each feature variable based on a rating criterion. Here, the model building unit (350) may calculate a final score by summing the scores assigned to each feature variable and determine the default rate of a specific company based on the final score.

[0087] In one embodiment, the model building unit (350) may classify the validation data so that a higher default rate corresponds to a higher grade based on the default rates of the learning data. For example, the model building unit (350) may assign a grade of 10 to the validation data with the highest default rate and a grade of 1 to the validation data with the lowest default rate during the process of classifying the validation data by applying a grade criterion to the validation data. In one embodiment, the model building unit (350) may sort the validation data in ascending or descending order based on the grades.

[0088] The default probability prediction unit (370) can predict the default probability of a specific company using a prediction model. Here, the prediction model may correspond to a default probability prediction model for each company learned based on learning data. The default probability prediction unit (370) can receive the final score of a specific company from the model construction unit (350) and determine the default rate of the specific company based on the final score. The default probability prediction unit (370) can display the final score on the user terminal (110), and here, the default probability prediction unit (370) can display the scores of the feature variables used in the process of deriving the final score and a description of the feature variables. In addition, the default probability prediction unit (370) can compare the default probabilities of each company by visually displaying the process of predicting the default probability of one or more companies based on a graph.

[0089] In one embodiment, the default probability prediction unit (370) may calculate a default variable rating and a final score based on the default probability output by the prediction model. Here, the default probability prediction unit (370) may perform classification based on the default variable rating assigned to each company during the process of predicting the default probability of one or more companies. Furthermore, the default probability prediction unit (370) may classify companies with a high probability of default into a risk group based on the default variable rating and display the companies on the user terminal (110).

[0090] The control unit (390) controls the overall operation of the corporate default probability prediction device (130), and the corporate default probability prediction device (130) can manage the control flow or data flow between the data collection unit (310), the feature variable extraction unit (330), the model construction unit (350), and the default probability prediction unit (370).

[0091]

[0092] Figure 4 is a flowchart explaining the functional configuration of the corporate default probability prediction device of Figure 1.

[0093] Referring to FIG. 4, a corporate default probability prediction device (130) can be implemented to execute a corporate default probability prediction service based on employment indicators according to the present invention.

[0094] Employment index data indicating the employment status for each company are collected through the data collection unit (310) (step S410). Here, the company-specific bankruptcy probability prediction device (130) collects employment index data on a monthly basis and can check whether the employment index data is continuous data for at least 12 months during the employment index data collection process. A plurality of feature variables for predicting the probability of bankruptcy are extracted from the employment index data through the feature variable extraction unit (330) (step S430). Here, the company-specific bankruptcy probability prediction device (130) can extract the number of national pension data for the current month, the number of new employees for the current month, the number of new employees for the current month, the number of employees for the current month, the amount of notice for the current month, the average salary, employment fluctuation tendency, MoM, YoY, and the Z-score for each mid-class industry as a plurality of feature variables for each company and store them in the database (150).

[0095] A prediction model for predicting the probability of default is sought by learning multiple feature variables through the model building unit (350) (step S450). Here, the corporate default probability prediction device (130) can construct a machine learning model that generates as output the continuous default probability of a specific company at each future point in time. In addition, the corporate default probability prediction device (130) can remove outliers of each feature variable and assign a score to each feature variable based on a rating standard set according to the distribution of the learning data. The default probability prediction unit (370) uses the prediction model to predict the probability of default of a specific company (step S470). Here, the corporate default probability prediction device (130) can calculate the default variable rating and final score based on the default probability output by the prediction model.

[0096]

[0097] FIG. 5 is a drawing illustrating one embodiment of a corporate default probability prediction device according to the present invention.

[0098] In Fig. 5, the corporate default probability prediction device (130) can collect corporate default-related feature variables based on the data collection unit (310). Here, the corporate default probability prediction device (130) can store the collected feature variables in a database (150) built based on a structured query language (SQL) and a non-relational database (Not Only Structured Query Language).

[0099] Next, the corporate default probability prediction device (130) can perform an exploratory data analysis (EDA) process through the model construction unit (350). Here, the corporate default probability prediction device (130) can receive each feature variable from the feature variable extraction unit (330) and use each feature variable as training data. The corporate default probability prediction device (130) can remove outliers in the training data based on the value of Z-score=3 for each feature variable and set a grade standard for each feature variable according to the distribution of the training data. Here, the corporate default probability prediction device (130) can classify the verification data by classifying the grade standard into 10 grades and applying it to the verification data. Here, the corporate default probability prediction device (130) can assign a grade of 10 to the verification data with the highest default rate and a grade of 1 to the verification data with the lowest default rate, and can assign a score of 1 to 10 to each feature variable according to the grade. The corporate default probability prediction device (130) can calculate a final score by collecting scores for each characteristic variable for a specific corporate.

[0100] Next, the corporate default probability prediction device (130) can receive the final score of a specific company based on the default probability prediction unit (370) and determine the default rate for the specific company based on the final score. The corporate default probability prediction device (130) can display the default rate of a specific company on the user terminal (110) and provide the score of each characteristic variable used in the calculated default rate and a description of each characteristic variable to the user terminal (110).

[0101]

[0102] Although the present invention has been described above with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various modifications and changes may be made to the present invention without departing from the spirit and scope of the present invention as set forth in the claims below.

[0103]

[0104] [Explanation of symbols]

[0105] 100: Employment Status Analysis System

[0106] 110: User terminal

[0107] 130: Employment Status Analysis Device

[0108] 150: Database

[0109] 310: Data Collection Department

[0110] 330: Feature Variable Extraction Unit

[0111] 350: Model Building Department

[0112] 370: Default probability prediction department

[0113] 390: Control Unit

Claims

1. Data collection department that collects employment indicator data indicating employment status by company; A feature variable extraction unit for extracting multiple feature variables for predicting the probability of default from the above employment index data; A model construction unit that learns the above multiple feature variables and pursues a prediction model for predicting the probability of default; and A device for predicting the probability of bankruptcy of a company based on employment indicators, comprising a bankruptcy probability prediction unit that predicts the probability of bankruptcy of a specific company using the above prediction model.

2. In paragraph 1, the data collection unit A device for predicting the probability of corporate bankruptcy based on an employment index, characterized by collecting the employment index data continuously aggregated for at least a specific period of time for each company.

3. In the first paragraph, the feature variable extraction unit A device for predicting the probability of corporate bankruptcy based on an employment index, characterized in that it extracts the number of national pension data for each company in the current month, the number of new employees in the current month, the number of resignations in the current month, the number of employees in the current month, the amount of notice for the current month, the average salary, employment fluctuation tendencies, MoM, YoY, and the Z-score by mid-class industry as the above multiple feature variables.

4. In paragraph 1, the model construction unit A device for predicting corporate default probability based on employment indicators, characterized in that it constructs a machine learning model as the prediction model, which receives the above multiple feature variables as inputs and generates the future continuous default probability of the company as outputs.

5. In the first paragraph, the model construction unit Remove outliers for each feature variable of the learning data, Based on the distribution of the above learning data, a rating criterion is set for each feature variable. Apply the above rating criteria to the verification data to classify the verification data, For each of the above characteristic variables, partial scores are given within a certain range based on the number of grades. A device for predicting the probability of corporate default based on employment indicators, characterized in that the final score is calculated by adding up the rating scores of each of the above multiple characteristic variables.

6. In paragraph 5, the model construction unit A device for predicting the probability of corporate default based on employment indicators, characterized in that the higher the default rate of the learning data, the higher the grade of the verification data is classified.

7. In paragraph 1, the default probability prediction unit A device for predicting corporate default probability based on employment indicators, characterized in that it calculates default variable grades and final scores based on the default probability output by the above prediction model.

8. In a default probability prediction method performed in a default probability prediction device, A step of collecting employment indicator data indicating employment status by company through the data collection department; A step of extracting multiple feature variables for predicting the probability of default from the employment index data through a feature variable extraction unit; A step of pursuing a prediction model for predicting the probability of default by learning the plurality of feature variables through a model building unit; and A method for predicting the probability of bankruptcy of a specific company based on employment indicators, comprising: a step of predicting the probability of bankruptcy of a specific company using the prediction model through a bankruptcy probability prediction unit;

Images