CN109102140A - A kind of growing quality evaluation method of base artificial intelligence and big data technology - Google Patents
A kind of growing quality evaluation method of base artificial intelligence and big data technology Download PDFInfo
- Publication number
- CN109102140A CN109102140A CN201810593866.6A CN201810593866A CN109102140A CN 109102140 A CN109102140 A CN 109102140A CN 201810593866 A CN201810593866 A CN 201810593866A CN 109102140 A CN109102140 A CN 109102140A
- Authority
- CN
- China
- Prior art keywords
- model
- enterprise
- growth
- roe
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Marketing (AREA)
- Educational Administration (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses the growing quality evaluation method of a kind of base artificial intelligence and big data technology, include the following steps: step 100, the data of selected part enterprise to establish the growth appraisal index model of enterprise；Step 200, model cross validation；Step 300, in the allowed band of the error of verification result, then growth appraisal index model is successfully established, otherwise, again the data of enterprise are chosen to establish growth evaluation index model, by establishing integrated study module and semi-supervised learning module on basic model, the precision of lift scheme that can be more perfect, stability and generalization ability, pass through integrated study, the prediction result of multiple models can be integrated, make the predictive ability of model more stable reliable, semi-supervised learning can make full use of the information of unmarked sample, further assurance model performance.
Description
Technical field
The present invention relates to valuation of enterprise technical field, specially a kind of business growth of base artificial intelligence and big data technology
Property evaluation method.
Background technique
In recent years in investment activity, Target Enterprise, the i.e. value of invested enterprise usually like looking at flowers in a fog, enable investment
It is indefinite that person ponders, then, the investment decision link mostly important at investor to the value assessment of Target Enterprise.Value assessment
Refer to that both parties buy or sell the value judgement made to target (equity or assets).Investor passes through certain method meter
The value of Target Enterprise is calculated, to provide basis of price for whether transaction is feasible.In investment process, the appraisal to Target Enterprise is
Whether it is successfully crucial.In terms of investor's angle, the either investor of shareholder or Target Enterprise, it is desirable to transaction value
Be conducive to one's own side.Since two-sided information is grasped in insufficient or subjective understanding there are deviation, cannot be fixed a price by a side and
Force at other side.At this moment it just needs that intermediary is engaged to make value assessment from the angle of economic technology.
It for medium-sized and small enterprises, either merges other enterprises or is merged by strong enterprises, be all its Fast Growth
Shortcut.And the value assessment to object is merged, just catering to the eager demand of the merger and acquisition market.
Growth is the core of medium-sized and small enterprises, and therefore, research and development is suitable for Medium and Small-sized High-tech Enterprises growth
The model and technology of evaluation reflect the growth situation of Medium and Small-sized High-tech Enterprises objective, just and soundly, for promoting middle-size and small-size high-tech
The development of skill business stability promotes sustained economic growth to have highly important theoretical and realistic meaning.
It is proposed to set a set of advanced, efficient automation small medium S&T enterprises growth in order to adapt to the needs in market
Investment value evaluation system, it is constant to promote evaluation accuracy, it reduces to artificial needs.
Existing growing quality assessment indicator system building verifying at present is insufficient, evaluation effect and result shortage are tested
It demonstrate,proves, the setting of weight is unreasonable in combination evaluation, and excessively relies on artificial judgment, there is the problems such as interference of subjective factor.
Summary of the invention
In order to overcome the shortcomings of that prior art, the present invention provide the enterprise of a kind of base artificial intelligence and big data technology
Growth appraisal method can be more perfect by establishing integrated study module and semi-supervised learning module on basic model
The precision of lift scheme, stability and generalization ability can integrate the prediction result of multiple models by integrated study, allow mould
The predictive ability of type is more stable reliable, and semi-supervised learning can make full use of the information of unmarked sample, further ensure mould
Type performance can effectively solve the problem of background technique proposes.
The technical solution adopted by the present invention to solve the technical problems is:
A kind of growing quality evaluation method of base artificial intelligence and big data technology, it is characterised in that: including walking as follows
It is rapid:
Step 100, the data of selected part enterprise establish the growth appraisal index model of enterprise；
Step 200, model cross validation；
In step 300, the allowed band of the error of verification result, then growth appraisal index model is successfully established,
Otherwise choosing the data of enterprise again to establish growth evaluation index model.
Preferably, the integrated study module and semi-supervised learning mould including basic model and foundation on basic model
Block；The historical data of enterprise is iterated operation by GBRT algorithm and obtains business data to be assessed by the basic model, will
The historical data of enterprise imitates having for expert estimation as unmarked sample, using business data to be assessed as machine learning is possessed
Marker samples；The historical data includes that enterprise uploads data and public data, and enterprise uploads data and public data and passes through specially
Family's marking, wherein the expert estimation obtains growth value point by GBRT algorithm；
Preferably, the integrated study module respectively obtains several marked samples pair by stacking Integrated Algorithm
The growth value point prediction result answered, all growth value point prediction results are fitted to obtain enterprise's totality by MLR algorithm
Growth value divides prediction result A；
Preferably, unmarked sample and marked sample are passed through Tri-training algorithm by the semi-supervised learning module
It makes full use of the information of unmarked sample to obtain new business data to be assessed, and obtains new growth value point prediction result B.
Preferably, the basic model includes 6 submodels, is respectively as follows: Predicting Technique innovation ability model M odel, in advance
Management ability model M odel is surveyed, predicts development potentiality model M odel, prediction management capability model Model, prediction operation ability
Model M odel and prediction profitability model M odel.
Preferably, submodel described in 6 is made of several corresponding factors of a model respectively, and by factor of a model
Determine each submodel weight shared in basic model.
Preferably, the model cross validation method of the step 200 is that other unmarked samples are used for the basis established
Model is forecast, and obtains growth value point, is then respectively obtained into according to integrated study module and semi-supervised learning module
Long value point prediction result A and growth value divide prediction result B, and the forecast of the growth value point of other unmarked samples is asked to miss
Difference records square adduction of prediction error.
Preferably, further includes: square adduction threshold value for setting prediction error is missed according to the forecast that model cross validation obtains
Square adduction of difference is to judge whether the growth appraisal index model of enterprise meets the requirements；
If square adduction of prediction error is within the threshold range, the growth appraisal index model of enterprise is conformed to
It asks；
If square adduction of prediction error is outside the range of threshold value, the growth appraisal index model of enterprise is not met
It is required that and reselecting the data of enterprise to establish the growth appraisal index model of enterprise；Until square of prediction error
Add within the threshold range.
Preferably, further include establishing economic model on the basis of basic model:
Definition: NI is net profit；
ROE is net assets income ratio；
B is profit retention ratio；
BV is net assets；
T and t-1 represent t and t-1；
Then the net profit of t-1, t can state are as follows:
NIt-1=BVt-1×ROEt-1；
NIt=﹙ BVt-1+b×NIt-1﹚ × ROEt；
Net profit growth rate g as a result,tIt is determined by following formula:
gt=NIt- NIt-1/NIt-1=ROEt- ROEt-1/ROEt-1+b×ROEt；
Work as ROEt=ROEt-1When, there is gt=b × ROE；
Meanwhile ROE can be unfolded are as follows:
ROE=ROA+D/E [ROA-i ﹙ 1-t ﹚]
Wherein: ROA is total assets return rate；D/E is debt book value/net assets；I is money rate；T is income tax
Rate；
In summary the economic model of growing quality can be obtained in formula:
G=b × ﹛ ROA+D/E [ROA-i ﹙ 1-t ﹚] ﹜.
Compared with prior art, the beneficial effects of the present invention are:
(1) present invention, can be more perfect by establishing integrated study module and semi-supervised learning module on basic model
Lift scheme precision, stability and generalization ability can be integrated the prediction result of multiple models, be allowed by integrated study
The predictive ability of model is more stable reliable, and semi-supervised learning can make full use of the information of unmarked sample, further ensures
Model performance；
(2) present invention effectively verifies the model of foundation using the method intersected with model, passes through until establishing
The model of verifying.
Detailed description of the invention
Fig. 1 is flow diagram of the invention；
Fig. 2 is basic model schematic diagram of the invention；
Fig. 3 is integrated study module diagram of the invention；
Fig. 4 is semi-supervised learning module diagram of the invention；
Fig. 5 is the fitting schematic diagram of GBRT algorithm of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
As shown in Figure 1, the present invention provides the growing quality evaluation method of a kind of base artificial intelligence and big data technology,
Include the following steps:
Step 100, the data of selected part enterprise establish the growth appraisal index model of enterprise；
Step 200, model cross validation；
In step 300, the allowed band of the error of verification result, then growth appraisal index model is successfully established,
Otherwise choosing the data of enterprise again to establish growth evaluation index model.
As shown in Figures 2 to 4, in the present embodiment, which includes basic model and builds
Found integrated study module and semi-supervised learning module on basic model；Wherein integrated study module and semi-supervised learning module
It is the two ways optimized to basic model.
In the present embodiment, its essence of growth appraisal index model is to learn expert by machine learning techniques to beat
Point, imitate expert to the decision process of enterprise valuation, establish assessment models, automatically and efficiently to upload data enterprise into
Row, the comprehensive parsing enterprise value of various dimensions, growth assessing model are with existing business data and corresponding expert estimation
Basic model, the public data crawled according to enterprise's upload data and crawler make growth appraisal to enterprise.
Six dimensions of basic model, which are set out, evaluates enterprise, is also equivalent to be divided into 6 submodels, be respectively as follows: pre-
Survey technology innovation ability model M odel (product technology or originality & innovation, advance and feasibility) predicts management ability model
Model (product development strategy reasonability, product profit mode, following positioning and development plan), predicts development potentiality model
Model (product market demand degree, core competitiveness, the difficulty of entrance), prediction management capability model the Model (group of foundation
Integrality, complementarity, stability and the industry background of team and resource integration capability), prediction operation ability model M odel (finance
Reasonability) and prediction profitability model M odel.
Evaluation logic, evaluation rule, the essential elements of evaluation of the growth assessing model store in systems.Evaluation procedure is being
It is completed in system, the evaluation output result of decision, for further auditing.
The historical data of enterprise is iterated operation by GBRT algorithm and obtains enterprise's number to be assessed by the basic model
According to being beaten using business data to be assessed as possessing machine learning and imitate expert using the historical data of enterprise as unmarked sample
The marked sample divided；The historical data includes that enterprise uploads data and public data, and enterprise uploads data and public data
By expert estimation, wherein the expert estimation obtains growth value point by GBRT algorithm；
GBRT algorithm has precision high, and generalization ability is strong, can handle nonlinear data, can handle multiple features type, logarithm
According to clean-up performance require it is relatively low the advantages that, be very suitable to growth assessing model needs summed data environment.
GBRT (gradientboostedregressiontrees) algorithm is by National Academy of Sciences academician
JH.Friedman is proposed, is a kind of Boosting algorithm that performance is extremely excellent.In practical applications, some algorithms have
Very good property, but exist simultaneously the poor disadvantage of Generalization Capability, such as decision tree, the relatively other algorithms of decision tree have with
Under several protrusions advantage: should be readily appreciated that and realize, the requirement to data prediction is lower, can handle data type and its simultaneously
Its General Properties.But often Generalization Capability is poor for the simple decision tree of structure, and Boosting is then that family can be by Generalization Capability
Poor weak learner is promoted to the algorithm of the excellent strong learner of Generalization Capability.The working mechanism of this race's algorithm be usually first from
Initial training collection trains a base learner, and the performance further according to base learner is adjusted training sample distribution, so that
The training sample that previous base learner does wrong more is paid close attention to subsequent, is then based on sample distribution adjusted to train
Next base learner, carries out repeatedly, until the learner quantity specified number in advance, finally by resulting study
Device is weighted combination.By Boosting algorithm, can be realized preferable while retaining base learner advantage as far as possible
Generalization Capability.
And the basic principle of GBRT is similar with common Boosting algorithm, using the simple regression tree model of structure as base
Plinth constructs new regression tree and model is added, the error of last round of model is corrected, more according to the fitting result of last round of model
After secondary iteration, a high-precision final mask is obtained by combining multiple regression tree models.But GBRT with it is common
The difference of Boosting algorithm maximum is that the calculating of GBRT each time can be reduced in residual error to reduce last residual error
A new model is established on gradient direction.In GradientBoost, each new model is in order to enable model before
Residual error is reduced toward gradient direction, and the way for increasing weighting with sample of the Boosting algorithm to mistake has very big difference.This
Sample, which is done, relative to the advantage of traditional Boosting algorithm to be that the efficiency of processing data can be obviously improved and effectively prevent intending
It closes, it is stronger to the resistance of abnormal point or extremum.
GBRT algorithm fit procedure by combining multiple simple regression trees of structure as shown in figure 5, obtained preferably quasi-
Close effect.
In the present embodiment, several marked samples are passed through stacking Integrated Algorithm point by the integrated study module
Corresponding growth value point prediction result is not obtained, and all growth value point prediction results are fitted by MLR algorithm is looked forward to
The growth value of industry totality divides prediction result A.
Integrated study module is capable of the precision of further lift scheme, stability and generalization ability.In marked sample
In the case where negligible amounts, may be simultaneously present multiple models has good performance on training set, and shows relatively
In actual use, the model performed better than on possible opposite training set has higher precision to the model of difference.Meanwhile sample size
In lesser situation, model is also easier to by extremum, and abnormal point etc. influences.By integrated study, multiple models can be integrated
Prediction result, make the predictive ability of model more stable reliable.
In the case where current training sample is less, it may be simultaneously present performance of multiple models on training set and approach,
As multiple performances can be obtained without significance difference by the way that different hyper parameters is arranged for the basic model GBRT selected by us
Different GBRT model, only choosing the model put up the best performance on training set in actual use and eliminating other models is clearly to have
Risk, it is possible to which preferably prediction effect can be obtained for actually using faced data instead by showing relatively poor model
Fruit.Specifically, since the hypothesis space of learning tasks is often very big, may there are multiple hypothesis to exist in terms of statistics
Reach equal performance on training set, if Generalization Capability may be caused bad because falsely dropping using single learner at this time, in conjunction with multiple
Learner can then reduce this risk；In terms of calculating, learning algorithm often falls into local minimum, some local poles
Generalization Capability corresponding to dot may be very bad, and can reduce the risk for falling into local minimum point by associative learning device
In terms of expression, vacation that the true hypothesis of certain learning tasks may not be considered in current learning algorithm
If then certainly invalid using single learner at this time in space, and by combining multiple learners, due to assuming that space has accordingly
Expanded, it is possible to which learn better approximation.Therefore, suitable combination strategy can be chosen, multiple models is combined, reduces wind
Danger, the stability of lift scheme entirety.And Stacking is then a kind of powerful Ensemble Learning Algorithms, can be used for combining multiple
GBRT model.
Stacking algorithm first goes out primary learner from initial training concentration training, then " generates " a new data set
For the secondary learner of training, in this new data set, the output of primary learner is taken as sample input feature vector, and initial
The label of sample is still taken as sample to mark.For growth assessing model, we choose GBRT as primary learner, choose
As secondary model, the output of comprehensive multiple GBRT obtains final at raising the price MLR (MultipleLinearRegression)
Value divides prediction result.
In the present embodiment, unmarked sample and marked sample are passed through Tri- by the semi-supervised learning module
Training algorithm makes full use of the information of unmarked sample to obtain new business data to be assessed, and obtains new growth value
Divide prediction result B.
Semi-supervised learning module uses Tri-training algorithm, makes full use of the information of unmarked sample.Growth is commented
Valence model imitates expert estimation by machine learning, but with submitting the enterprise of data to increase, possess expert estimation has label
The ratio of the relatively unmarked sample of sample is necessarily smaller and smaller, to influence the performance of model.Traditional modeling method training pattern
Without using unmarked sample, but in fact, bulk information is equally existed in unmarked sample can be used for training pattern, semi-supervised
Habit can make full use of these information, further assurance model performance.
With the increase for the enterprise for uploading data, the quantity of unmarked sample necessarily will be much larger than having by expert estimation
The quantity of marker samples by the information for being included using unmarked sample rather than only relies on marked sample training pattern,
It is capable of the Generalization Capability of further lift scheme, for example, although unmarked sample does not directly include mark information, but if they
It is from same Data Source Independent with profile samples with marked sample, then the letter about data distribution that they are included
Breath obviously contributes to establish model.The semi-supervised learning algorithm that growth assessing model uses belongs to for Tri-training
In the method (disagreement-basedmethods) based on disagreement, using multiple learners, by between multiple learners
Disagreement utilize Unlabeled data, since growth assessing model will train multiple learners in integrated study module, because
This Tri-training after sample size is enough will be embedded in growth assessing model well.
Co-training is initially for " multiple view " (multi-view) design data.In many practical applications, one
A data object often possesses multiple property sets simultaneously, and each property set just constitutes a view.Co-training is initial
Foundation is the compatible complementarity using multiple view, it is assumed that data have multiple abundant and conditional sampling views, can use a letter
Single method utilizes Unlabeled data: a classifier is respectively trained out based on marked sample first on each view, so
After allow each classifier to select oneself " most safe " unmarked sample respectively to assign pseudo- label, and pseudo- marker samples are provided
To other classifiers as newly-increased marked sample for training renewal learning device, then iteration carries out this process.Although side
Method is simple, but existing theoretical proof, if view sufficiently and conditional sampling, passes through Co-training using unmarked sample
The Generalization Capability of Weak Classifier is promoted to any height.Also, in the ungratified situation of conditional independence, pass through Co-
Training, performance still can have a degree of promotion, in this embodiment it is not even necessary to which data possess multiple view, only need between base learner
There are significant disagreements (or difference) can pass through Co-training improving performance.
The basic model includes 6 submodels, is respectively as follows: Predicting Technique innovation ability model M odel, and energy is managed in prediction
Power model M odel predicts development potentiality model M odel, prediction management capability model Model, prediction operation ability model M odel
With prediction profitability model M odel.
Submodel described in 6 is made of several corresponding factors of a model respectively, and is determined respectively by factor of a model
A submodel weight shared in basic model, composition table are as follows.
Predicting Technique innovation ability model M odel
| Model1 | Variables |
| V1 | Belong to the enterprise in national high and new technology industrial development zone |
| V2 | Belong to the enterprise in national university science and technology park |
| V3 | Place industry field |
| V4 | Registered capital |
| V5 | Directly it is engaged in research and development person's number |
| V6 | More than junior college scientific and technological member's number |
| V7 | Researcher's ratio |
| V8 | The above employee's ratio of junior college |
| V9 | Patent holds number |
| V10 | Belong to domestically leading level |
| V11 | Domestic market status ranking |
| V12 | Current year main business income |
| V13 | Current year main business cost |
| V14 | Current year main business tax |
| V15 | Current year income from main operation rate |
| V16 | Management level number |
| V17 | Master's number |
| V18 | Doctor's number |
| V19 | Average age |
Predict management ability model M odel
Predict development potentiality model M odel
Prediction management capability model Model
| Model4 | Variables |
| V1 | Belong to the enterprise in national high and new technology industrial development zone |
| V2 | Belong to the enterprise in national university science and technology park |
| V3 | Place industry field |
| V4 | Registered capital |
| V5 | Paid-up capital |
| V6 | Employee's sum |
| V7 | Directly it is engaged in research and development person's number |
| V8 | More than junior college scientific and technological member's number |
| V9 | Researcher's ratio |
| V10 | The above employee's ratio of junior college |
| V11 | Patent holds number |
| V12 | Belong to domestically leading level |
| V13 | Domestic market status ranking |
| V14 | Management level number |
| V15 | Master's number |
| V16 | Doctor's number |
| V17 | Average age |
Predict operation ability model M odel
Predict profitability model M odel
| Model6 | Varibles |
| V1 | Belong to the enterprise in national high and new technology industrial development zone |
| V2 | Belong to the enterprise in national university science and technology park |
| V3 | Place industry field |
| V4 | Registered capital |
| V5 | Paid-up capital |
| V6 | Employee's sum |
| V7 | Main business income |
| V8 | Main business income prediction |
| V9 | Occupation rate of market |
| V10 | Income from main operation rate |
| V11 | Net assets |
| V12 | The item number of research and development |
The model cross validation method of the step 200 is, the basic model that other unmarked samples are used to establish into
Row forecast, and growth value point is obtained, growth value is then respectively obtained according to integrated study module and semi-supervised learning module
Point prediction result A and growth value divide prediction result B, and ask the prediction error of the growth value point of other unmarked samples, note
Record square adduction of prediction error.
Set square adduction threshold value of prediction error, square adduction of the prediction error obtained according to model cross validation come
Judge whether the growth appraisal index model of enterprise meets the requirements；
If square adduction of prediction error is within the threshold range, the growth appraisal index model of enterprise is conformed to
It asks；
If square adduction of prediction error is outside the range of threshold value, the growth appraisal index model of enterprise is not met
It is required that and reselecting the data of enterprise to establish the growth appraisal index model of enterprise；Until square of prediction error
Add within the threshold range.
It additionally provides in the present embodiment, the economic model established on the basis of basic model:
Definition: NI is net profit；
ROE is net assets income ratio；
B is profit retention ratio；
BV is net assets；
T and t-1 represent t and t-1；
Then the net profit of t-1, t can state are as follows:
NIt-1=BVt-1×ROEt-1；
NIt=﹙ BVt-1+b×NIt-1﹚ × ROEt；
Net profit growth rate g as a result,tIt is determined by following formula:
gt=NIt- NIt-1/NIt-1=ROEt- ROEt-1/ROEt-1+b×ROEt；
Work as ROEt=ROEt-1When, there is gt=b × ROE；
Meanwhile ROE can be unfolded are as follows:
ROE=ROA+D/E [ROA-i ﹙ 1-t ﹚]
Wherein: ROA is total assets return rate；D/E is debt book value/net assets；I is money rate；T is income tax
Rate；
In summary the economic model of growing quality can be obtained in formula:
G=b × ﹛ ROA+D/E [ROA-i ﹙ 1-t ﹚] ﹜.
The economic model of growing quality shows that net profit growth is by the coefficient knot of variable b, ROA, D/E, i and t
Fruit.In these variables, some variables be it is controllable, some variables be it is uncontrollable, and some variables be not exclusively it is controllable.
Wherein, income tax rate t apparently pertains to uncontrolled variable, but its influence to all enterprises is all consistent, and a state
Family and area income tax rate be usually to maintain in one section of longer time stablize it is constant, therefore, can in economic model
It is regarded as constant；Controlled variable includes b and D/E, and the two reflects the dividend policy and financing obstacle of enterprise respectively, and these political affairs
The final of plan puts into effect the strategic decision and managerial ability depending on the financial situation of enterprise itself and manager-level again；Not exclusively may be used
Controlling variable includes ROA and i, for ROA, is not only limited by the spy of the enterprises such as managerial ability, the scale of enterprise itself
Sign factor will also directly be influenced by industry organization's structure and macroeconomy situation；And for i, not only by country
The influence of the uncontrollable element such as monetary policy, inflation rate, meanwhile, will also be by the result of Corporate finance policy --- capital
The influence of structure.According to the above analysis, by the model, not only there is enterprise controllable we can see that influencing the factor of growth
Internal factor, further include the external factor such as the uncontrollable economic development of enterprise, industrial structure.
It, being capable of small enterprise in objective evaluation using reasonable evaluation method by establishing the growth of SMEs evaluation model
The state of development of industry and its constructional aspect of place industry, scientifically evaluate the growth of SMEs, predict the following hair
Exhibition trend.
One, government administration section is assisted to be better understood by enterprise and park development situation.
Enterprise comments house that government administration section can be helped to obtain more comprehensive enterprise operation Information of Development, to administrative area state-owned enterprise
Industry development has accurate assurance, so as to reasonably formulate sme development support policy, carries out reasonable
Industrial pattern.
Two, Economy Garden administrative department is assisted to make rational planning for garden and service enterprise.
Garden administrative department is using the management state for looking forward to commenting house that can in real time, comprehensively hold enterprise in garden, thus section
The rich potential enterprise of introduction eliminates backward enterprise, provides decision basis for Garden Planning and Government supports.It is more preferable simultaneously
The demand of understanding enterprise, provide better service for enterprise in garden.
Three, a kind of appraisal tool is provided for financial institution, realizes diversification risk resolution.
Enterprise comments house to the holographic portraits of medium-sized and small enterprises so that financial institution can be more accurately to the growth of medium-sized and small enterprises
Potentiality make assessment, more good investment target are quickly filtered out from a large amount of medium-sized and small enterprises, by locking after further examining
It sets the goal, preferably guarantees investment return, prevention and control risk.Enterprise comments house to enable financial institution from a large amount of poor benefit repeatability works
It frees in work and is engaged in the work of high benefit value, promote its business fast-developing.
Four, it assists enterprise to find problem of management, promotes company governance and science decision is horizontal.
Enterprise comments house to evaluate by the analysis of six big dimensions growing quality, assists enterprise to find out and is hidden in deep layer
Main problem, so that it carries out specific aim improvement.Be conducive to enterprise to mark industry overall development status, specify itself locating competition
There is the developing state of current and future in position and clearly recognizes, and is conducive to enterprise and makes reasonable Developing Decision.Meanwhile it looking forward to
It comments the evaluation result of family by the approval of financial institution and government garden, enterprise can be helped to carry out financing expansion and seek garden
The policy of area and government administration section is helped.
It is obvious to a person skilled in the art that invention is not limited to the details of the above exemplary embodiments, Er Qie
In the case where without departing substantially from spirit or essential attributes of the invention, the present invention can be realized in other specific forms.Therefore, no matter
From the point of view of which point, the present embodiments are to be considered as illustrative and not restrictive, and the scope of the present invention is by appended power
Benefit requires rather than above description limits, it is intended that all by what is fallen within the meaning and scope of the equivalent elements of the claims
Variation is included within the present invention.Any reference signs in the claims should not be construed as limiting the involved claims.
Claims (9)
1. the growing quality evaluation method of a kind of base artificial intelligence and big data technology, characterized by the following steps:
Step 100, the data of selected part enterprise establish the growth appraisal index model of enterprise；
Step 200, model cross validation；
In step 300, the allowed band of the error of verification result, then growth appraisal index model is successfully established, otherwise
If, choose the data of enterprise again to establish growth evaluation index model.
2. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 1, special
Sign is: including basic model and establishing integrated study module and semi-supervised learning module on basic model；The base
The historical data of enterprise is iterated operation by GBRT algorithm and obtains business data to be assessed by plinth model, by the history of enterprise
Data are as unmarked sample, using business data to be assessed as the marked sample for possessing machine learning imitation expert estimation；
The historical data includes that enterprise uploads data and public data, and enterprise uploads data and public data passes through expert estimation,
Described in expert estimation by GBRT algorithm obtain growth value point.
3. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 2, special
Sign is: several marked samples are respectively obtained corresponding growth by stacking Integrated Algorithm by the integrated study module
Value divides prediction result, and all growth value point prediction results are worth by the growth that MLR algorithm is fitted to obtain enterprise's totality
Divide prediction result A.
4. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 2, special
Sign is: the semi-supervised learning module makes full use of unmarked sample and marked sample by Tri-training algorithm
The information of unmarked sample obtains new business data to be assessed, and obtains new growth value point prediction result B.
5. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 2, special
Sign is: the basic model includes 6 submodels, is respectively as follows: Predicting Technique innovation ability model M odel, and energy is managed in prediction
Power model M odel predicts development potentiality model M odel, prediction management capability model Model, prediction operation ability model M odel
With prediction profitability model M odel.
6. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 5, special
Sign is: submodel described in 6 is made of several corresponding factors of a model respectively, and is determined by factor of a model each
Submodel weight shared in basic model.
7. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 1, special
Sign is: the model cross validation method of the step 200 is that the basic model for being used to establish by other unmarked samples carries out
Forecast, and growth value point is obtained, growth value point is then respectively obtained according to integrated study module and semi-supervised learning module
Prediction result A and growth value divide prediction result B, and ask the prediction error of the growth value point of other unmarked samples, record
Square adduction of prediction error.
8. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 7, special
Sign is, further includes: square adduction threshold value for setting prediction error, according to square for the prediction error that model cross validation obtains
Adduction is to judge whether the growth appraisal index model of enterprise meets the requirements；
If square adduction of prediction error is within the threshold range, the growth appraisal index model of enterprise meets the requirements；
If square adduction of prediction error is outside the range of threshold value, the growth appraisal index model of enterprise, which is not met, to be wanted
It asks, and reselects the data of enterprise to establish the growth appraisal index model of enterprise；Until prediction error square plus
Within the threshold range.
9. the growing quality evaluation method of a kind of base artificial intelligence and big data technology according to claim 2, special
Sign is, further includes establishing economic model on the basis of basic model:
Definition: NI is net profit；
ROE is net assets income ratio；
B is profit retention ratio；
BV is net assets；
T and t-1 represent t and t-1；
Then the net profit of t-1, t can state are as follows:
NIt-1=BVt-1×ROEt-1；
NIt=﹙ BVt-1+b×NIt-1﹚ × ROEt；
Net profit growth rate g as a result,tIt is determined by following formula:
gt=NIt- NIt-1/NIt-1=ROEt- ROEt-1/ROEt-1+b×ROEt；
Work as ROEt=ROEt-1When, there is gt=b × ROE；
Meanwhile ROE can be unfolded are as follows:
ROE=ROA+D/E [ROA-i ﹙ 1-t ﹚]
Wherein: ROA is total assets return rate；D/E is debt book value/net assets；I is money rate；T is income tax rate；
In summary the economic model of growing quality can be obtained in formula:
G=b × ﹛ ROA+D/E [ROA-i ﹙ 1-t ﹚] ﹜.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810593866.6A CN109102140A (en) | 2018-06-11 | 2018-06-11 | A kind of growing quality evaluation method of base artificial intelligence and big data technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810593866.6A CN109102140A (en) | 2018-06-11 | 2018-06-11 | A kind of growing quality evaluation method of base artificial intelligence and big data technology |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109102140A true CN109102140A (en) | 2018-12-28 |
Family
ID=64796795
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810593866.6A Pending CN109102140A (en) | 2018-06-11 | 2018-06-11 | A kind of growing quality evaluation method of base artificial intelligence and big data technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109102140A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097216A (en) * | 2019-04-11 | 2019-08-06 | 企家有道网络技术(北京)有限公司 | For the prediction technique and device of enterprise development, server |
| CN110619462A (en) * | 2019-09-10 | 2019-12-27 | 苏州方正璞华信息技术有限公司 | Project quality assessment method based on AI model |
| CN112734271A (en) * | 2021-01-19 | 2021-04-30 | 建信金融科技有限责任公司 | Growth curve regression model training method and enterprise evaluation index generation method |
| CN113450009A (en) * | 2021-07-06 | 2021-09-28 | 北交金科金融信息服务有限公司 | Method and system for evaluating enterprise growth |
| CN115081950A (en) * | 2022-07-28 | 2022-09-20 | 江西省智能产业技术创新研究院 | Enterprise growth assessment modeling method, system, computer and readable storage medium |
| TWI795809B (en) * | 2021-06-17 | 2023-03-11 | 華南商業銀行股份有限公司 | Business evaluation system and method therefore |
| TWI820731B (en) * | 2021-10-29 | 2023-11-01 | 美商萬國商業機器公司 | Electronic system, computer-implemented method, and computer program product |
-
2018
- 2018-06-11 CN CN201810593866.6A patent/CN109102140A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110097216A (en) * | 2019-04-11 | 2019-08-06 | 企家有道网络技术(北京)有限公司 | For the prediction technique and device of enterprise development, server |
| CN110619462A (en) * | 2019-09-10 | 2019-12-27 | 苏州方正璞华信息技术有限公司 | Project quality assessment method based on AI model |
| CN112734271A (en) * | 2021-01-19 | 2021-04-30 | 建信金融科技有限责任公司 | Growth curve regression model training method and enterprise evaluation index generation method |
| TWI795809B (en) * | 2021-06-17 | 2023-03-11 | 華南商業銀行股份有限公司 | Business evaluation system and method therefore |
| CN113450009A (en) * | 2021-07-06 | 2021-09-28 | 北交金科金融信息服务有限公司 | Method and system for evaluating enterprise growth |
| TWI820731B (en) * | 2021-10-29 | 2023-11-01 | 美商萬國商業機器公司 | Electronic system, computer-implemented method, and computer program product |
| CN115081950A (en) * | 2022-07-28 | 2022-09-20 | 江西省智能产业技术创新研究院 | Enterprise growth assessment modeling method, system, computer and readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109102140A (en) | A kind of growing quality evaluation method of base artificial intelligence and big data technology | |
| Phelps | Structural slumps: The modern equilibrium theory of unemployment, interest, and assets | |
| Akintoye et al. | Policy, finance & management for public-private partnerships | |
| Banker et al. | Some models for estimating technical and scale inefficiencies in data envelopment analysis | |
| Chen et al. | A comprehensive model for fuzzy multi-objective portfolio selection based on DEA cross-efficiency model | |
| CN109726905A (en) | A kind of method and system of enterprise value portrait evaluation | |
| CN105931116A (en) | Automated credit scoring system and method based on depth learning mechanism | |
| DE112013006007T5 (en) | Simulation of institutions | |
| Eyal-Cohen | Legal mirrors of Entrepreneurship | |
| Livsey | Economic Diversification Through A Knowledge-Based Economy In the United Arab Emirates: A Study of Progress Toward Vision 2021 | |
| Lempert et al. | Transportation Planning for Uncertain Times: A Practical Guide to Decision Making Under Deep Uncertainty for MPOs | |
| Ibukun-Falayi et al. | Improving human resource accounting through international financial reporting standards | |
| Arroyo et al. | Implementing a three-level balanced scorecard system at Chilquinta Energia | |
| French | Building rural community resilience through innovation and entrepreneurship | |
| Berg et al. | Scenario Building for Development Cooperation–Methods Paper | |
| Thamprasert et al. | Simulated trial and error experiments on productivity | |
| Rohani et al. | Prioritising (ranking) of indexes for measuring intellectual capital using FAHP and fuzzy TOPSIS techniques | |
| Sanyal | India: decentralised planning: themes and issues | |
| Adebayo | Taxation and economic diversification as tools for a sustainable economy: lessons from Canada | |
| Abe | The future of big data analysis for the Philippine’s rice industry using the integration of mind mapping and future wheel in scenario building | |
| Riepina et al. | Risks of Agrobusiness Digital Transformation | |
| Manzi | The impact of the Zambian economy (2015–2017) on the sustainability of the construction industry | |
| Jev et al. | Economic Policies and Challenges of Development in Nigeria | |
| Gremillion | AN ANALYSIS OF CAPITAL FLOW TO CONSTRUCT AN ECONOMIC TYPOLOGY IN A COMPARATIVE CASE STUDY OF THREE ANCIENT CITIES: THIRD KINGDOM OF UR, NEW KINGDOM OF THEBES, AND CLASSICAL ATHENS | |
| Demuth | Special Feature" Socio-Economic Role of Islamic Finance and |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181228 |