KR100776187B1 - Method of technology evaluation - Google Patents

Abstract

A technology evaluation method is provided to enhance performance of a technology evaluation model and to reflect systematically a score recognized via a knowhow of an evaluator by performing a small adjustment in an interval with a higher estimated correctness and performing many adjustments in an interval with a lower estimated correctness among a confidence interval of post probability on nonpayment. A technology evaluation method comprises the following several steps. Probability calculated according to a Logit model is scored and a risk level is given to evaluation target technology in order to reflect possibility of nonpayment on the evaluation target technology(S123). A technology level is given to the evaluation target technology according to a score calculated a weighting score model in order to reflect a technology feature, a business feature and a market feature of the evaluation target technology(S125). The risk level and the technology level are integrated on an x axis and a y axis of a matrix(S129). A step of obtaining the risk level from a Logit score and an environment score includes the following several steps. A variable of a technology evaluation score factor is calculated. A variable of an economic index factor is calculated. A company state variable is calculated. A Logit model inputting value is determined by using a basic calculated value from each step. The Logit score is calculated by inputting the variable of a technology evaluation score factor into a Logit function 1 and calculating excellence probability. The environment score is calculated by inputting the variable of an economic index factor and the company state variable into a Logit function 2 and calculating excellence probability.

Description

Method of technology evaluation

1 is an explanatory diagram illustrating the concept of a technology evaluation method according to a preferred embodiment of the present invention;

2 is a flowchart for calculating an evaluation grade for each technical evaluation item according to the present invention;

3 is a flowchart for calculating a risk score (logit model) for obtaining a risk level according to the present invention;

4 is a flowchart of a method of predicting management performance;

5 is a screen example illustrating a drawing part and a commentary part of the KTCP matrix;

6 is a flowchart for building a KTCP matrix;

7a is a KTCP matrix diagram showing a technical evaluation grade;

7b is a KTCP matrix diagram showing management performance prediction;

7c is a KTCP matrix diagram illustrating technicality vs. business feasibility;

7D is a KTCP matrix diagram illustrating technicality vs. marketability;

7E is a KTCP matrix diagram illustrating marketability vs. business feasibility;

7F is a KTCP matrix diagram illustrating financial safety vs. technology business feasibility;

8A is a reception management screen of the technology evaluation acceptance step,

8b is a screen for inputting the details of the reception of the technical evaluation acceptance step;

8C is an input screen of an enterprise overview,

8d is a representative status input screen;

8E is a financial statement input screen,

8F is an economic environment indicator data management screen;

8g is an evaluation cover cover creation screen,

8H is a screen for inputting 45 data to create an evaluation table,

8i is a rating item rating screen of Example 1 of measurement item;

8J is a rating item rating screen of Example 2 of the metering item;

8K is a rating item rating screen of a check item example;

8L is a rating item rating screen of an example of a rating item;

8m is a screen showing a balance matrix,

8n is a screen showing a result of rating,

8o is an evaluation result display step, a screen showing a result of calculating a technical evaluation grade, performance analysis, and survival probability (= 1-default probability);

Figure 9a is a graph comparing the predictive power of the business unit between the conventional technology evaluation method and the technology evaluation method according to the present invention that does not reflect the risk level of the technology,

Figure 9b is a ROC graph comparing the predictive power of the business unit between the conventional technology evaluation method that does not reflect the risk level of the technology and the technology evaluation method by the cognitive scoring model according to the present invention.

The present invention relates to a technology evaluation method that reflects the characteristics of the target technology, the nature of the company having the technology, and the economic and environmental characteristics faced by the company for use in the field of technology finance, such as credit, credit guarantee, investment, etc. It is possible to predict the probability of default and management performance by year according to the grade.

In general, there are two types of technology evaluation, technical evaluation and technology value evaluation. Technical evaluation can be divided into technology rating evaluation and technology evaluation. First of all, technology rating evaluation is a type of evaluating the feasibility of technology in terms of percentage or rating. Technology evaluation is an evaluation of the overall technical capability of a company. It comprehensively evaluates individual skill levels, technical personnel, intellectual property rights, and research facilities. And technology valuation is a type that converts the value of individual technology financially to calculate the price of technology transaction or the collateral value of intellectual property.

Here, a rating model, a profile model, a checklist model, and the like are used as the rating method of the technology. A scoring model is a method of assigning a rating to each evaluation item of technicality, and applying a weight based on the importance between evaluation items. Profile model expresses evaluation elements in figures and has the advantage of quickly identifying the advantages and disadvantages of the technology. Checklist Model sets Check Points to be reviewed, Cut-Off Criteria for each Checklist, and then checks whether all Checklists meet certain criteria That's how.

The conventional technology valuation model is a model that determines each valuation element and assigns a point (importance) to each valuation element and evaluates it by a rating or a grade. There was a limit to having.

In addition, the conventional technology evaluation method does not reflect the risk of bankruptcy of the company due to the economic environment or economic factors, so the reliability of the evaluation was not high.

Recently, technical evaluation methods have been proposed to improve these problems. However, since the technical evaluation items depend heavily on the evaluator's supervision, distorted phenomena may occur, and the results of the technology score model established based on these evaluation results are also distorted. Can be.

In general, the sub-assessment items used in the scoring method are largely composed of objective items that are evaluated equally regardless of the evaluators, and subjective items that are highly subject to change depending on the evaluators. In the evaluator's evaluation behavior, in many cases, the evaluator perceives the composite score a priori prior to evaluating the individual items. Thereafter, individual scores are adjusted for subjective sub-assessments other than objective evaluation items that have less freedom of evaluation, and scores corresponding to a priori overall recognition scores are calculated. Therefore, in many cases, the individual scores are insignificant, whereas the composite scores are significant, and this characteristic is the biggest advantage of the scoring method. Under these circumstances, if the relationship between individual item scores and accidents is grasped and the item weights are obtained, there is a possibility that an incorrect scoring model is derived. Therefore, an evaluation model that reflects the characteristics of objective and subjective evaluation items and a priori overall recognition score is required.

In addition, the conventional technology evaluation method does not consider various types of accidents (ie overdrafts, natural bodies, bad credit, etc.) in showing the change in corporate defaults over time, and simply does the same. Considering bankruptcy shows a limit that does not reflect its importance.

Under these circumstances, in July 2004, the government announced the Comprehensive Measures to Strengthen the Competitiveness of SMEs. Accordingly, an improved form of technology evaluation model was introduced to evaluate the feasibility of more advanced technology projects, and a rating scheme considering the default rate of technology commercialization. There is a need to improve the credible technical grade and certification system for various purposes such as construction.

The present invention has been made to solve the above problems, an object of the present invention is to provide a technology evaluation method reflecting the environmental risk and technology commercialization risk according to the economic environment and economic factors.

In addition, another object of the present invention is to reconstruct mutually exclusive technical evaluation items based on literature review and the results of the significance analysis of past technical evaluation indicators, and to introduce objection, neutrality, It is to improve the complementarity to provide a technology evaluation method that has proven reliability in the evaluation index and application method.

It is also an object of the present invention to provide a technology evaluation method that can show the change in corporate default over time in consideration of various causes of corporate default.

It is also an object of the present invention to provide a technology evaluation method that can predict the performance of a new company by analyzing the technology evaluation cases of similar companies in the past.

In addition, the object of the present invention is to provide conditions for the activation of technical financial support for innovative leading SMEs by providing reliable technology evaluation certification grades for various purposes such as technology transfer, technology transaction and technology finance (financing and investment). It is to provide a technique evaluation method that can be used.

In addition, the present invention uses the distribution of cognitive total scores to reflect the cognitive total scores of evaluators rather than independently evaluating individual technical scores considered as input variables in the process of calculating technical commercial risk scores to obtain risk ratings. It is to provide a technique evaluation method that can improve the performance of the evaluation model by improving the evaluation model to adjust the contrast with the confidence interval.

In addition, the present invention can be applied not only to the company's rating evaluation by improving the life data analysis model so that it is possible to grasp the various causes of corporate defaults through the competition risk model and to identify the default risk level and predict the survival probability. In other words, it is possible to predict the survival time and risk probability by reflecting the characteristics of bankrupt companies, and provide a technology evaluation method that can help to identify the cause of an accident.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

In order to achieve the above object, the present invention comprises the steps of: obtaining a risk rating by assigning a rating by rating the probability calculated by the logit model to reflect the possibility of business division of the technology to be evaluated; Obtaining a technology grade that gives a grade according to the grade calculated by the weighted grade model in order to reflect the technicality, businessability, and marketability of the technology to be evaluated; And obtaining a technical evaluation certification level that gives the rating by combining the risk level and the technical level by a matrix having X and Y axes, and calculating a risk score for the risk level as logit score and environment. From ratings,

The calculating of the logit score and the environmental score may include calculating a technical evaluation score factor variable; Calculating economic indicator factor variables; Calculating an enterprise state variable; Determining a logit model input value based on a basic output value obtained at each of the steps; Calculating a logit score by inputting the technical evaluation score factor variable into logit function 1 and calculating a probability of a superior probability; Including the economic indicator factor variable and the company state variable into the logit function 2 and calculating the environmental probability by calculating the excellent probability,

Injecting the factor variable into the logit function, calculating the logit score, and calculating the probability of probability, the confidence interval of the probability probability calculated by the logit model to reflect the total score of the evaluator is properly compared with the distribution of the total score of the evaluator. It provides a technology evaluation method characterized in that for calculating the updated excellent probability, calculating the logit score using the updated excellent probability.

In particular, an updated logistic regression equation defined as Equation (E) below is used to adjust the distribution of cognitive total scores by contrasting the confidence interval of the superior probability calculated from the logit model while reflecting the cognitive total scores of the evaluator. It is characterized by.

(E)

(E)

here,

: 업데이트된 사고확률,

Updated accident probability,

: 평가자에 의한 인지총점,

: Total cognitive score by evaluator,

: 인지총점

의 평균,

: Cognitive total score

Mean,

,

: 각각 인지총점

의 최대값과 최소값,

,

: Cognitive total score for each

The maximum and minimum values of,

,

: 각각

의 신뢰구간의 하한과 상한.

,

: each

And upper bound of the confidence interval.

는 객관적인 평가항목의 벡터,

는 각 평가항목에 대응되는 회귀계수

의 추정된 계수값임)(

Is the vector of objective endpoints,

Regression coefficient corresponding to each evaluation item

Estimated coefficient of)

In addition, the technical evaluation method of the present invention further comprises the step of calculating the default probability of default according to the grade through the competition risk model,

Calculating the probability of default for each year includes calculating a linear combination of input variables and coefficients, and calculating survival probability (= 1-default probability) over a given time;

In calculating the survival probability, the survival probability is predicted and calculated using a competitive risk model in consideration of various characteristics of the company and various accident causes.

Hereinafter, preferred embodiments of the technical evaluation method according to the present invention will be described in detail with reference to the accompanying drawings.

1 is an explanatory view illustrating the concept of a technology evaluation method according to a preferred embodiment of the present invention, the technology evaluation method according to the present invention is largely to obtain a risk level (S123), to obtain a technology level (S125), Comprising the step (S129) of obtaining the technology evaluation certification grade, the step of obtaining the default probability of default according to the grade (S133), and the management performance prediction step (S135), are divided below.

The evaluation grade of the technical evaluation method according to the present invention is classified into a risk grade (S123), a technical grade (S125), and a technical evaluation (certification) grade (S129).

The risk level (S123) is a statistical analysis of the commercialization conditions, technical characteristics, and related matters of the business to be commercialized or in progress, to review the possibility of the business unit based on the technology, and at the same time, the economic environment that may affect the business promotion of the company. As a technology-based enterprise risk rating that combines factors, etc., the business unit of the target company (technology) is rated by a certain criterion by rating the probability of good or poor calculated by the logit model to reflect the risk. 10 grades (aaa-d) are given.

Technology grade (S125) is a technology-based grade that calculates the possibility of business success and technology innovation capability of a company based on the technology by comprehensively examining the technology, business feasibility, and marketability of the technology to be commercialized or in progress. In order to reflect the technical, business, and marketability of), ratings are given by applying certain criteria to the ratings calculated by the weighted rating model, and displayed as 10 grades (V1 to V10).

Technical evaluation (certification) grade (S129) is a comprehensive review of the risk grade and technical grade by the matrix (Matrix), and given a grade, and expressed as 10 grades (AAA ~ D).

This rating system is shown in Table 1 below.

Here, the risk level (S123) and the technical level (S125) are given by certain criteria according to the risk score and the technical score calculated by each model, and the technical evaluation (certification) grade (S129) is the risk grade and the technical grade. Is a grade given by comprehensively reviewing the matrix using X and Y axes, and the results are shown in Table 2 below.

Hereinafter, a method for obtaining a risk level, a technology level, and a technology evaluation certification level by the technology evaluation method according to the present invention will be described in detail.

The technical evaluation items used to obtain the risk level, technology level, and technology evaluation certification level in this technology evaluation method are shown in Table 3 below. The technical evaluation items have a hierarchical structure of large, medium, small items, and detailed review items. Based on the evaluation results of the detailed review items, scores (grades) for each technical evaluation item (small items) are calculated. The scores of (sub-items) are used as inputs for the calculation of the risk and technical grades.

2 is a flowchart for calculating an evaluation grade for each technology evaluation item according to the present invention. As shown in FIG. 2, the technology evaluation table is logged in (S201), and the evaluation contents of the detailed review items for each technology evaluation item (small items) are input ( S203).

The evaluation contents to be input are classified into three categories, weighing evaluation items, check evaluation items, and evaluator evaluation items, respectively, according to the evaluation method.

Weighing evaluation items are items that are evaluated by the values automatically calculated by the system using separately inputted company data such as company overview data and company financial data. This is an item that is automatically calculated by the system, so it is very important to check for missing data.

Weighing evaluation items are divided into evaluation items based on the size (level) of the calculated value itself and evaluation items based on the level of the industry average value (average ratio by industry, etc.).

The calculation method for each item is shown in Table 4 below.

Check evaluation items are evaluated automatically by the number of items checked by the evaluator by checking the check items presented as evaluation criteria.

The status of check evaluation items is shown in Table 5 below.

The subjects of the evaluator's evaluation items are evaluation items other than the measurement and check evaluation items, and the evaluator assigns and evaluates an arbitrary level according to the evaluation criteria presented on the pop-up.

The evaluator evaluation method for each evaluation item is summarized in Table 3 above.

After inputting the evaluation contents of the detailed review items for each technical evaluation item (small item) by the evaluation method as described above (S203), check whether the input item is missing (S205), and if there is a missing item, input it ( S207).

Once all the detailed review items have been entered, the score (grade) for each technical evaluation item (small item) is automatically generated by the 5 × 5 Balance Matrix or the 5 × 5 × 5 Balance Matrix according to the number of detailed review items for each technical evaluation item (small item). It is calculated (S209).

 An example of calculation by the 5 × 5 Balance Matrix is shown in Table 6 below.

In addition, an example of calculation by the 5 × 5 × 5 Balance Matrix is shown in Table 7 below.

3 is a flowchart for calculating a risk score (logit model) for obtaining a risk level according to the present invention. The risk prediction model establishes a model by applying the technical evaluation data evaluated by the technical evaluation table to the logit model.

As a procedure for obtaining a logit score for obtaining a risk level, it is first checked whether an evaluation input of a detailed review item for each technical evaluation item (small item) is missing (S301). This is a step of automatically checking whether the evaluation item is missing a score or not, or whether other input variables are missing. If a check result is missing an input of an evaluation score, the item is supplemented (S303).

Next, using the scores (grades) of each evaluation technology input item (small items), the technical evaluation score factor variable is calculated using a function expression (S305).

Factor variable calculation is shown in Table 8 below.

       n = 1, 2, ... 16)

When the calculation of the factor variable is finished (S305), next, the economic indicator factor variable is calculated by using a function formula using separately stored economic indicators (S307).

The list of economic indicator variables is shown in Table 9 below.

이다.Here, the economic indicator factor variable calculation formula is

to be.

: 경제요인변수(n=1,2,3,4)(

: Economic factor variable (n = 1,2,3,4)

(t기의 y번째 변수의 표준화) =

(normalization of yth variable of t) =

: n번째 경제요인변수에 대한 t기의 y번째 경제변수의 요인계수)

: Factor factor of the y th economic variable of t for the n th economic factor variable)

Next, the company state variable is calculated (S309).

The list of variables for calculating the company state variable is shown in Table 10 below, which is obtained by separately inputted company data.

Since the company status variable is an item in which an input error or a missing entry always exists during the regular survey and data management DB input process, the list of company status variables should be output on the screen (Pop-UP) to be reviewed by the evaluator before the final rating is calculated. It constitutes (S311).

Next, the basic calculation value and the logit model input value are determined and stored (S313).

Subsequently, the technical evaluation score factor variable is input to the logit model to obtain a logit score 1 (S315), and a probability probability is calculated (S317) to obtain a logit score (S319).

The items and variable names for calculating the logit score are shown in Table 11 below.

Here, in estimating the probability of goodness using the logit model, the present invention has a main feature in reflecting the cognitive total points inferred from the evaluator's experience and know-how.

In general, the sub-assessment items used in the scoring method are largely composed of objective items that are evaluated equally regardless of the evaluators, and subjective items that are highly subject to change depending on the evaluators. In the evaluator's evaluation behavior, in many cases, the evaluator perceives the composite score a priori prior to evaluating the individual items. Thereafter, individual scores are adjusted for subjective sub-assessments other than objective evaluation items that have less freedom of evaluation, and scores corresponding to a priori overall recognition scores are calculated. Therefore, in many cases, individual scores are significantly less significant, whereas composite scores are significant, which is the biggest advantage of the scoring method. Under these circumstances, if the relationship between individual item scores and accidents is grasped and the item weights are obtained, there is a possibility that an incorrect scoring model is derived. Therefore, an evaluation model that reflects the characteristics of objective and subjective evaluation items and a priori overall recognition score is required.

To this end, the present invention performs logistic regression analysis using non-accident relations using only objective evaluation items evaluated similarly according to evaluators. In other words, this model excludes the a priori influence of the evaluator.

First, the equation of the logistic regression analysis is shown in the following equation (1).

(1)

(One)

는 객관적인 평가항목의 벡터를 말하며,

는 각 평가항목에 대응되는 회귀계수이다. 또한 종속변수로서 사고일 경우 Y=1이고 비사고일 경우 Y=0이다. F( )는 다음의 식(2)와 같은 로짓함수를 말한다.Where P is the probability of default

Is a vector of objective endpoints,

Is the regression coefficient corresponding to each evaluation item. The dependent variable is Y = 1 in case of accident and Y = 0 in case of non-accident. F () is a logit function as shown in Equation (2).

(2)

(2)

는

의 추정된 계수값이고, 이를 이용하여 부도확률의 추정값

는 다음의 식(3)과 같이 추정된다.here

Is

Is an estimated coefficient value of, and the estimated value of default probability

Is estimated by the following equation (3).

(3)

(3)

의 신뢰구간은 다음의 식(4)와 같이 구할 수 있다.On the other hand, at the confidence level of 100 (1-α)

The confidence interval of can be calculated by the following equation (4).

(4)

(4)

는 표준정규분포를 말하고,

는

의 추정된 표준오차를 말한다. 더불어

과

는

의 신뢰구간의 하한과 상한을 말한다.here

Refers to the standard normal distribution,

Is

The estimated standard error of. together

and

Is

The lower and upper bound of the confidence interval.

In case of having a posterior probability around 0 or 1 due to the characteristics of the logistic regression model, the model has a high discrimination ability for the evaluation result, and the confidence interval in this section is very narrow. On the contrary, in case of having a posterior probability of around 0.5, the interval of discrimination is low and the confidence interval is very wide. In other words, in consideration of these characteristics, the evaluator's a priori overall recognition score is reflected by the following equation, and in the model analyzed above, it is small in the section with high discrimination and large in the section where it is not. A technology evaluation model can be established that can be combined appropriately. This is shown in the following equation (5).

(5)

(5)

은 업데이트된 사고확률을 말하며,

는 평가자에 의한 인지총점이고,

는 인지총점

의 평균을 말한다. 한편,

과

은 각각 인지총점

의 최대값과 최소값을 말한다.here

Is the updated probability of accident,

Is the total cognitive score by the evaluator,

Cognitive total score

Says the mean. Meanwhile,

and

Cognitive total score

The maximum and minimum values of.

의 비율에 의해 사후확률을 업데이트하여 주게 된다. In general, the scorecard by the evaluator has a characteristic of not having a minimum value of 0, and in order to take this into account, the range of the perceived total score (Max (z) -Min (z)) and

The posterior probability is updated by the ratio of.

In Eq. (5), the evaluator's cognitive total score is adjusted based on the average of cognitive total scores. The estimated default probability is the expected value of default probability, which reflects the characteristics that contrast with the average of the total scores.

에 의해 산출하며(S317), '로짓평점 = 업데이트된 우량확률 × 100'의 식에 의해 계산한다(S319).The updated Score is

It is calculated by (S317), and calculated by the formula of 'Logit score = updated rainfall probability × 100' (S319).

In conclusion, this adjustment method adjusts the distribution of the evaluator's cognitive total score against the confidence interval of the posterior default probability (or rainfall probability), so that less adjustment is made in the high accuracy probability range, while many in the low accuracy probability interval. Coordination will be reflected systematically.

Next, the items and variable names for calculating the environmental scores are as shown in Table 12 below, and the logit score 2 is calculated by injecting the economic indicator factor variables and the company state variables into the logit model (S321), and obtaining a good probability (S321). S323) to obtain an environmental score (S325).

+ 환경평점 ×

'의 산식에 의하여 위험평점을 구한다(S327). 여기서

는 로짓평점과 환경평점을 가중결합하기 위한 가중치이다. Once logit and environmental scores are found, 'risk rating = logit rating ×

+ Environmental rating ×

Calculate the risk score by the formula of (S327). here

Is a weight for weighted combining logit and environmental scores.

When the risk score is obtained, as shown in FIG. 1, the risk level is calculated accordingly (S123).

The following is a step of obtaining the technical grade by the weighted rating model (S125). Here, as shown in Table 13 below, a method of obtaining technical scores by technical evaluation items (small items) is illustrated.

The weight of each evaluation item reflects the characteristics of each industry and applies the weights by different industries.

When the technical score is determined by the weighted rating model, the technical grade is assigned by the evaluation criteria of Table 14 below (S125).

Once the risk grade (S123) by the logit model and the technical grade (S125) by the weighted rating model are calculated, the step of finally assigning the technical evaluation certification grade by the matrix evaluation criteria will be described (S129).

Calculation of the technology evaluation certification grade is based on the evaluation by the matrix as shown in Table 2 above.

The following describes the steps of calculating the default probability of failure (= 1-survival probability) according to the grade by the life data analysis model (survival analysis model) (S133). This step is to calculate the default probability of failure (= 1-survival probability) according to the technical evaluation certification grade by using the evaluation result of the technical evaluation table as an input variable.

Lifetime (survival) data consists of a period from a certain starting point to the occurrence of an event, for example, from the time of credit guarantee (financing) according to the technical evaluation and evaluation results to the failure of the business (failure to commercialization). This time is called survival.

The life data analysis (survival analysis) analyzes the survival period, calculates the probability and risk of survival after the time point of the technology evaluation and credit guarantee (financing) company, and measures the life expectancy of the company to be evaluated. Provides additional information such as trends in business default rates by grade.

However, in the analysis of general life data, there is a disadvantage in that it does not show the degree of risk in consideration of the relationship between various causes of corporate defaults (natural bodies, defaults, bad credit, etc.).

In the step of calculating the survival probability, the present invention establishes a competitive risk model based on the Weibull distribution, wherein the scale parameter of the Weibull distribution is composed of a regression model in consideration of input variables reflecting the borrower's personality. It is characterized by modeling such that the shape parameter (shape parameter) of the Weibull distribution is different for each cause of competition risk.

를 와이블 분포(Weibull distribution)를 가정한 것이다. Equation (6) below shows how long it takes for firm i to go bankrupt because of j

Is assumed for the Weibull distribution.

(6)

(6)

와

는 각각 척도모수와 형상모수를 의미한다. 이러한 분포의

와

는 각각 확률밀도함수와 누적확률밀도함수이며, 이를 이용하여 생존함수를 구하면 다음의 식(7) ~ (9)와 같다. Where i (i = 1, ..., n) refers to individual companies and j ^th (j = 1, ..., k) refers to the causes of accidents that result in bankruptcy,

Wow

Denotes scale parameter and shape parameter, respectively. Of this distribution

Wow

Are the probability density function and the cumulative probability density function, respectively, and the survival function is calculated using the following equations (7) ~ (9).

(7)

(7)

(8)

(8)

(9)

(9)

분위수 수명은 식 (10)과 같다.Where i is the probability of survival for the cause j

The quantile life is given by equation (10).

(10)

10

는 기업사고 원인별로 차이가 있으며, 또 다른 모수

는 기업 i의 특성변수(

)를 반영하기 위하여

로 처리하여 회귀모형화 한다면 다음의 식(11)로 표현할 수 있다. Where one parameter of the Weibull distribution

Varies by cause of business accidents, and another parameter

Is the characteristic variable of

To reflect)

If the regression model is processed by the following equation, it can be expressed by the following equation (11).

(11)

(11)

와 어떠한 원인에 의해서든 사고가 발생할 확률인

를 다음과 같이 구할 수 있다.By using Equation (8), the probability that an accident does not occur by any accident cause

And the probability of an accident

Can be obtained as

(12)

(12)

(13)

(13)

Equation (12) can be used to calculate the probability of survival of firm i, and the probability of default of firm i can be found from equation (13).

By performing this method, it is possible to predict not only the survival probability for each accident cause but also the survival probability considering all accidents, and can be applied for various purposes.

Table 15 below lists the input parameters for calculating the survival probability of the life data analysis model.

Table 16 below is an example of shape parameters and time setting values for calculating the annual survival probability. In this case, the time is set to one month as 12 months in units of months.

Given these variables, the probability of default (1-survival probability) at time t is

① calculate the linear combination of inputs and coefficients,

② Calculate the probability of default (= 1-survival probability) at a given time point t.

An example of a default probability (= 1-survival probability) estimation result at a given time point t is shown in FIG. 8O.

The following describes the step (S135) of predicting management performance by case-based reasoning (S121).

Case-Based Reasoning (CBR) is a method that finds existing cases that are similar to new cases and uses similarities to the solved problems of existing cases. It extracts past cases of technology evaluation and analyzes the performance of companies with similar technology evaluation cases.

4 is a flowchart illustrating a method of predicting management performance, and a step-by-step calculation method will be described with reference to the flowchart.

First, a rating for each technical evaluation index is calculated (S401). Here, the rating calculated by the technical evaluation table is used as it is.

Next, Euclidean distance is measured (S403).

Euclidean distance is an indicator of the similarity measure. It is calculated as the square root of the sum of the squares of the distances between the technical evaluation sub-item ratings and the comparable companies (past grading cases).

: 비교대상기업(과거 평점사례)의 평점

: Rating of the comparable company (past rating cases)

: 평가대상기업의 평점

: Rating of the subject company

: 가중치

: weight

                          n = 1, 2,... , 16

※ The Euclidean distance is calculated for each of the comparable companies (past grading cases).

Next, the Euclidean distance is measured to extract 10 cases with the minimum minimum distance, that is, the highest similarity (S405), and extract the growth, profitability, and activity of the 10 companies (S407).

The results of performance evaluation by case-based reasoning for the companies to be evaluated are accumulated in the companies to be compared (past grading cases) and used as the companies to be compared. In the above, the risk rating, technology rating, technology evaluation certification grade, default probability according to grade, and management performance prediction according to the present invention were described.

When the results of the technical evaluation according to the technical evaluation method of the present invention described above are analyzed from various viewpoints such as risk level, technical performance, and business feasibility, and displayed in a two-dimensional diagram (hereinafter referred to as 'KTCP matrix'), the utilization thereof may be increased. Let's take a look at how to make KTCP matrix and its utilization.

5 is an exemplary screen illustrating a drawing part and an explanation part of the KTCP matrix. As shown, the KTCP matrix is a technology evaluation and consulting matrix that expresses the characteristics of a company as a chart, with the X- and Y-axis as the technology, business feasibility, technology business, and risk class variables used in the technology evaluation model, respectively. It is composed of wealth and commentary.

The chart section classifies the technical performance, business feasibility, and risk perspective evaluation results of the companies to be evaluated, and displays them in a matrix.

The commentator is in charge of the function of automatically outputting a description of each matrix after combining and configuring the matrix from various viewpoints.

The technology used as an indicator of the KTCP matrix comprehensively evaluates technology development promotion capability, investment and infrastructure for technology development, technology innovation, completeness, and scalability. Comprehensive evaluation of commercialization capability and marketing ability.

6 is a flowchart for constructing a KTCP matrix, in order to construct a KTCP matrix, a variable selection step (S601), a risk rating, a technology rating, a financial stability rating calculation step (S603), a technology, a business feasibility, a technology and a business feasibility rating step (S605), the section setting step of the rating for each variable (S607), the matrix table creation step (S609), the matrix diagram section construction step (S611) is completed.

In the variable selection step (S601), each variable of the technical evaluation risk level, technical level, and technical variable is selected. The technical evaluation risk level uses a technology-based risk level calculated by the technology evaluation model, and the technology level is a rating or a rating calculated by the scoring method for all evaluation items of the above-mentioned technical evaluation index. .

The technical variable is composed of technical evaluation items of the technical evaluation index, as shown in Table 17 below.

Marketability variables are composed of marketability evaluation items of the technology evaluation index as shown in Table 18 below.

Table 19 below shows the feasibility ① variables.

Table 20 below is the feasibility ② variable.

Technological feasibility variables consist of the technical and feasibility evaluation items of the technical evaluation index as shown in Table 21 below.

The financial stability variable uses the financial grade using the financial model of the corporate credit rating system as the financial stability variable.

When the variable selection step is completed, a risk grade, a technical grade, and a financial stability grade (grade) are calculated (S603).

The risk level is a technology-based risk level calculated by the above-described technology evaluation model.

The technical grade is a grade calculated by the above-described technical evaluation model.

Financial stability is the financial grade produced by the financial model of the corporate credit rating system.

Next, the technical, business, and technical business scores are calculated (S605).

According to the scoring model by the scoring method method, the score for each field (variable) is converted to 100 points according to the following formula.

When the grade is calculated, in order to construct a chart, first, each grade (grade) for each variable is classified into six sections and a section is set (S607).

Next, a matrix table is created using the calculated score values (S609).

Finally, the matrix diagram portion and commentary portion are constructed (S611).

7A is a KTCP matrix diagram showing the technical evaluation grade, which is prepared with the technical grade as the X axis and the hazard grade as the Y axis. The technology evaluation grade evaluates the feasibility of commercialization on the basis of technology, technology development status, and commercialization ability in relation to the target technology. Means that the possibility of default is assessed by the risk assessment model.

7B is a KTCP matrix diagram showing management performance prediction. In the case of technology-based businesses, especially small and medium-sized companies, it is difficult to predict future management performance in comparison with traditional companies, and based on this, various management performance cases of technology companies are analyzed. To predict. Management performance is calculated based on four financial indicators, including profitability, growth, stability, and activity. The higher the grade, the better the performance.

FIG. 7C is a KTCP matrix diagram describing technicality vs. business performance, and is a chart in which the technical content, technical level, and the like of the target technology, commercialization capability, marketing situation, and future profit outlook are prepared on the matrix.

FIG. 7D is a KTCP matrix diagram illustrating technicality vs. marketability, and is a chart in which the technical content of a technology, the technology level, and the like are compared to the market size, growth prospects, and competition conditions on a matrix.

FIG. 7E is a KTCP matrix diagram illustrating marketability vs. business feasibility, and is a chart in which the market size, growth prospects, competition status, technology commercialization capability, marketing status, and future profit outlook of the technology are prepared in a matrix.

FIG. 7F is a KTCP matrix diagram illustrating financial safety vs. technology business feasibility, which is a chart that expresses the future growth prospects (technical feasibility) such as technology, business feasibility and profitability of the technology and the short-term financial stability at the time of evaluating the company against the matrix.

The following is a screen on a computer which actually applies the technology evaluation method according to the present invention described above, and each screen will be described according to the evaluation procedure of the present invention.

8A is a reception management screen of the technology evaluation acceptance step, and FIG. 8B is a reception detail input screen of the technology evaluation reception step.

8C to 8F are input stages of corporate data, and FIG. 8C is an input screen of an enterprise overview (industry, business history, registration, externality, venture company, etc.), FIG. 8D is a representative status input screen, and FIG. 8E is a financial statement. 8F is an economic environment indicator data management screen.

8G to 8N are evaluation data input steps, FIG. 8G is a screen for creating an evaluation cover sheet, FIG. 8H is a screen for inputting 45 pieces of data to create an evaluation table, and FIG. 8I is a screen for assigning a rating item in Example 1 of measurement item. 8J is an evaluation item rating granting screen of the example of the measurement item 2, FIG. 8K is an evaluation item rating granting screen of the example of the check item, FIG. 8L is an evaluation item rating granting screen of the example of the evaluator evaluation, and FIG. 8M is a balance FIG. 8N is a screen showing a matrix, and FIG. 8N is a screen showing a rating calculation result.

8o is an evaluation result display step, which is a screen showing a technology evaluation grade, performance analysis, and survival probability (= 1-default probability).

In addition, Figure 9a is a graph comparing the predictive power of the business unit between the conventional technology evaluation method and the technology evaluation method according to the present invention that does not reflect the risk level of the technology, Figure 9b is a conventional technology that does not reflect the risk rating of the technology The business unit between the evaluation method and the technology evaluation method using the cognitive score model according to the present invention is a ROC graph comparing the predictive power.

According to FIG. 9A, the conventional technology evaluation method in which the risk level of the technology is not reflected should show a higher default rate as the downgrade is not. However, according to FIG. 9B, the comparison of the performance evaluation by the area under the ROC graph is also low. have. However, according to the present invention, the higher the grade, the lower the default rate, and the performance evaluation by the area under the ROC graph also shows an improved result compared to the conventional technology evaluation method.

As described above, all preferred embodiments of the technology evaluation method according to the present invention are realized according to a program in which the present invention is implemented by DB data from a computer.

As described above, the technical evaluation method according to the present invention has the following effects.

1) Technology commercialization to obtain risk rating In establishing a rating model that calculates a risk score, the evaluator considers the fact that individual scores are reassigned from the perceived total score by the evaluator's supervision rather than independently evaluated. It is to improve the valuation model so that the distribution of cognitive total scores is adjusted against the confidence interval of the posterior probability of individual firm defaults created from statistical modeling. In this way, a small adjustment can be made in the section with high prediction accuracy among the confidence intervals of the posterior probability for bankruptcy, while many adjustments can be systematically reflected in a section with low prediction accuracy, improving the performance of the technology evaluation model and evaluator. It is possible to evaluate the technology that systematically reflects the cognitive score based on the know-how.

2) Improved life cycle data analysis model to identify the degree of default risk over time by reflecting various types of accidents related to technology financing (eg, natural bodies, bankruptcy, bad credit, etc.) through competitive risk model Therefore, it is possible not only to apply the technology to the commercialization of technology in institutions for the purpose of credit, credit guarantee, investment, etc., but also to predict the survival time and risk probability by reflecting the characteristics of the bankrupt company. To help.

Claims (3)

In the technology evaluation method using a terminal comprising a microprocessor and a storage device, the method, Obtaining a risk level by assigning a rating to the probability calculated by the logit model by using the terminal to reflect the possibility of business division of the technology to be evaluated; Obtaining a technical grade that gives a rating according to the rating calculated by the weighted rating model using the terminal to reflect the technicality, businessability, and marketability of the technology to be evaluated; And using the terminal to obtain a technical evaluation certification level which gives the rating by combining the risk level and the technical level by a matrix having X and Y axes, and a risk score for obtaining the risk level. Is calculated from logit and environmental scores, The calculating of the logit score and the environmental score may include calculating a technical evaluation score factor variable; Calculating economic indicator factor variables; Calculating an enterprise state variable; Determining a logit model input value based on a basic output value obtained at each of the steps; Calculating a logit score by inputting the technical evaluation score factor variable into logit function 1 and calculating a probability of a superior probability; Including the economic indicator factor variable and the company state variable into the logit function 2 and calculating the environmental probability by calculating the excellent probability, Injecting the factor variable into the logit function, calculating the logit score, and calculating the probability of probability, update the confidence interval of the probability probability calculated by the logit model to reflect the total score of the evaluator according to the distribution of the total score of the evaluator. Calculating the excellent probability, and calculating the logit score using the updated excellent probability. The method according to claim 1, Technical evaluation method characterized by using the updated logistic regression equation defined as Equation (E) below so that the distribution of cognitive total points can be adjusted by contrast with the confidence interval of the posterior probability while reflecting the cognitive total score of the evaluator. .

(E) here,

Updated accident probability,

: Total cognitive score by evaluator,

: Cognitive total score

Mean,

,

: Cognitive total score for each

The maximum and minimum values of,

,

: each

And upper bound of the confidence interval. (

Is the vector of objective endpoints,

Regression coefficient corresponding to each evaluation item

Estimated coefficient of) The method according to claim 1, After the step of obtaining the technical evaluation certification level to give a rating by combining the risk level and the technical level by the matrix using the X-axis and Y-axis using the terminal, Computing the default probability of default according to the grade by using a competitive risk model using the terminal, Calculating the probability of default for each year includes calculating a linear combination of input variables and coefficients, and calculating survival probability (= 1-default probability) over a given time; And calculating the survival probability using the competitive risk model considering various characteristics of the company and various causes of accidents.

Images