CN109214598A - Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk - Google Patents
Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk Download PDFInfo
- Publication number
- CN109214598A CN109214598A CN201811249353.XA CN201811249353A CN109214598A CN 109214598 A CN109214598 A CN 109214598A CN 201811249353 A CN201811249353 A CN 201811249353A CN 109214598 A CN109214598 A CN 109214598A
- Authority
- CN
- China
- Prior art keywords
- risk
- model
- residential quarters
- collateral
- grading
- 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
- 238000000034 method Methods 0.000 title claims abstract description 38
- YHXISWVBGDMDLQ-UHFFFAOYSA-N moclobemide Chemical compound C1=CC(Cl)=CC=C1C(=O)NCCN1CCOCC1 YHXISWVBGDMDLQ-UHFFFAOYSA-N 0.000 title claims abstract description 23
- 238000013480 data collection Methods 0.000 claims abstract description 10
- 238000004422 calculation algorithm Methods 0.000 claims description 29
- 238000004458 analytical method Methods 0.000 claims description 15
- 238000013278 delphi method Methods 0.000 claims description 4
- 238000010835 comparative analysis Methods 0.000 claims description 2
- 238000009826 distribution Methods 0.000 claims description 2
- 238000005259 measurement Methods 0.000 claims description 2
- 241001269238 Data Species 0.000 claims 1
- 238000013481 data capture Methods 0.000 claims 1
- 239000011159 matrix material Substances 0.000 abstract description 10
- 238000011156 evaluation Methods 0.000 abstract description 5
- 238000004445 quantitative analysis Methods 0.000 abstract description 4
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 238000012502 risk assessment Methods 0.000 description 12
- 238000005311 autocorrelation function Methods 0.000 description 9
- 238000004451 qualitative analysis Methods 0.000 description 5
- 206010019133 Hangover Diseases 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000003064 k means clustering Methods 0.000 description 3
- 238000013178 mathematical model Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000012360 testing method Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000005520 cutting process Methods 0.000 description 2
- 238000007418 data mining Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 238000000513 principal component analysis Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 241001123248 Arma Species 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 230000012010 growth Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000002948 stochastic simulation Methods 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- 238000001744 unit root test Methods 0.000 description 1
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
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- 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
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/16—Real estate
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Technology Law (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses the batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk, includes the following steps: that residential quarters collateral risk grading Connotation Definition, residential quarters collateral risk grading risks and assumptions identification, data collection, different factor models are built, the adjustment of total model buildings, model and model are issued.To realize the residential quarters collateral risk grading that can be quantified, seek in batches.The present invention is a kind of quantitative analysis method of multi-specialized combination, real estate macroscopic view, microcosmic two dimensions are preferably fused together, and pass through GIS tool auxiliary judgment, utilize space matrix, risk rating rule is made by mathematical principle, so that each urban cells Risk Evaluation standard in the whole nation is unitized, standardization.It grades to each cell mass in national different cities, grading range covering surface is big, and speed is fast, to help the monitoring high-risks cell such as financial institution, takes precautions against high risk visitor group.
Description
Technical field
The present invention relates to Real Estate Finance and information analysis to combine field, especially a kind of by data mining, real estate
Valuation technique, applied mathematics, GIS are combined, and are based primarily upon K-MEANS clustering and ARIMA model to predict residential quarters
The batch ranking method of collateral risk.
Background technique
The more traditional real estate risk analysis of academic circles at present research is based on qualitative analysis more, even if there is a small number of risks
Judgement is quantitative analysis, is also focused mostly in systemic financial risks judgement, the analysis of real-estate market overall risk, such as:
Using a variety of mathematical models to Chinese real-estate market carry out risk assessment, for estate investment projects risk assessment,
General financial risks evaluation etc.；It is chiefly used in the risk judgment of macroscopic aspect or State-level for the quantitative grading of risk,
Risk judgment research rarely is carried out to microstructure layer grade.
Therefore following shortcoming and defect can be caused respectively:
1, traditional risk analysis, defines risk intension range and relatively obscures, and determines risk also without unified rule.Due to right
The standard difference that the understanding of risk intension is different, determines risk, causes analysis result can be with stronger personal colors, no
The Risk Results determined with analysis personnel same subject matter can also show differentiation.
2, traditional quantitative risk rating focuses mostly in general financial risks judgement, real-estate market overall risk point
Analysis, the residential mortgage risk rating of rare microcosmic point.It is qualitative for providing the main method of collateral risk analysis at present for bank
Analysis, such as: the mortgage public lecture for being supplied to bank carries out qualitative analysis to collateral risk by the following aspects: (1) mortgaging
The market liquidity of real estate and quick cashability；(2) fluctuation tendency of real estate Fair Market Value is mortgaged；(3) it predicts
The overall price situation and variation tendency in market, comprehensive descision evaluate the risk status of object.But shortage quantifies collateral risk
Method.
3, traditional risk analysis artificially judges that accounting is larger.The analysis personnel for being engaged in microcosmic point qualitative analysis are because a
The limitation of people's experience, not high to the identification of risks and assumptions, the data surface contacted is narrower, sees without global, also without the overall situation
Date comprision, risk analysis result is often more unilateral, and can not carry out risk to cell each in entire city in batches
Analysis.And the analysis personnel of macroscopic aspect qualitative analysis are engaged in, and compare and be absorbed in global risk analysis, such as State-level,
Or city level, lacking the risk judgment of microcosmic aspect, this has resulted in the isolating property in traditional risk analysis dimension,
Effectively the risk analysis of macroscopic aspect can not be combined with microcosmic point, so that traditional risk analysis globality is poor.
Therefore traditional Risk Evaluation lacks more targeted risk definition and judgment criteria, and globality is poor, and
It is more to depend on subjective experience, less Dependent Algorithm in Precision and auxiliary tool.
Summary of the invention
In order to solve the above technical problem, the present invention provides one kind to be based primarily upon K-MEANS clustering and ARIMA prediction
The batch ranking method of the residential quarters collateral risk combined, can preferably solve following problems:
1, the present invention defines clearly risk intension range, and can solving risk, to define judgment criteria caused by disunity inconsistent
Problem, facilitating analysis personnel has unified module to particular risk.
2, the present invention is the quantitative analysis method combined based on appraisal of real estate, data mining, mathematical model, GIS, is had
Relatively comprehensive sturdy theoretical basis can be well solved the risk being easy to appear based on qualitative analysis and judge by accident or because of experience problem
Caused by erroneous judgement etc. technical problems.
3, the present invention can help financial institution, and the house class solved as unit of cell especially for business bank's batch is supported
It gives as security credit risk quantitatively to grade problem, improves business bank to the risk identification ability of house class house property.
The technical scheme adopted by the invention to solve the technical problem is that:
Based on the batch ranking method of K-MEANS and ARIMA model prediction residential quarters collateral risk, include the following steps:
Step 1: defining residential quarters collateral risk grading intension.
Step 2: identification residential quarters collateral risk grading risks and assumptions
The present invention is based on Real Estate Appraising Theories, carry out real estate risks and assumptions using related algorithm in Delphi method and statistics
Judgement, determine influence real estate risk Main Factors after, and then identify residential quarters collateral risk grading main wind
The dangerous factor.
Step 3: data collection
The residential quarters collateral risk factor obtained according to step 2 collects fingers at different levels by analyzing the intension of each risks and assumptions
Mark data.Data collection is using web crawlers and the method manually combined.
Step 4: different algorithm models is built according to different risks and assumptions
By the data collection of step 3, the related data of different stage index is obtained, while being built according to different types of index
Corresponding subalgorithm model.
Such as: cell real factor part index number algorithm: clustering (K-Means, SpectralClustering)；
Market fluctuation predicted portions index algorithm: ARIMA, GM (1,1) etc.；
Locational factor part index number algorithm: Nemerow Index method etc..
Step 5: completing model buildings
Completion model buildings: first, by the calculating of all kinds of submodels of step 4, obtain each risks and assumptions in two-level index
Score value；Second, Raw performance weight is determined, in conjunction with Delphi expert graded by algorithm with regress analysis method, determines every grade
The final score value of risk indicator weight；Third calculates the total score of each cell according to determining index score value and weighting weight；
4th, the Statistical Distribution of indexs at different levels is analyzed, and combine clustering algorithm, evaluates the risk class of each cell.
Step 6: seeking National Residential cell risk rating result in batches
According to the model that step 5 is completed, grade to National Residential cell batch.
Step 7: adjustment model
For step 6 obtain risk rating as a result, providing model evaluation by the expert of real estate different field as a result, comparison
Analysis, adjusting and optimizing model parameter so that result be really more nearly.
Step 8: the publication of National Residential cell collateral risk rating model
The measurement model of risk rating is issued, while carrying out product introduction.
The invention has the advantages that
The present invention is a kind of quantitative analysis method of multi-specialized combination, compared with traditional qualitatively method, operational analysis speed with
Rating result precision has great promotion.
The present invention has more sturdy theoretical basis, can preferably be fused together real estate macroscopic view, microcosmic two dimensions, and
By GIS tool auxiliary judgment, using space matrix, risk rating rule is made by mathematical principle, makes each city in the whole nation
Cell Risk Evaluation standard is unitized, standardizes.
The present invention can grade to each cell mass in national different cities, and grading range covering surface is big, and speed is fast, to
The monitoring high-risks cells such as financial institution are helped, high risk visitor group is taken precautions against.
Detailed description of the invention
Invention is further explained with reference to the accompanying drawing.
Fig. 1 is present invention assessment residential quarters collateral risk grading flow chart.
Specific embodiment
As shown in Figure 1, it includes: residential quarters collateral risk grading intension circle that the present invention, which assesses the step of real estate risk,
Fixed, residential quarters collateral risk grading risks and assumptions identification, data collection, difference factor model builds, total model buildings, model
Adjustment and model publication, thus realize can residential quarters collateral risk that is quantitative, seeking in batches grade.
The implementation method of each step is illustrated below with reference to Fig. 1.
Residential quarters collateral risk grading Connotation Definition
According to " Real Estate Appraising Theory and method ", " real estate mortgage appraisal instruction ", " business bank's security management refers to
Draw " etc. files, the intension boundary of risk rating is defined, residential quarters collateral risk is graded is defined as: Residential Area
Price fluctuation risk and realization risk assessment during property mortgage.
Identify residential quarters collateral risk grading risks and assumptions
The present invention is based on Real Estate Appraising Theories, and the judgement of real estate risks and assumptions is carried out using Delphi method, collects a large amount of rooms
Expert in terms of real estate and financial field is to the opinion of residential quarters collateral risk grading risks and assumptions, and comprehensive each expert is to each
The opinion of risks and assumptions, and by the principal component analysis in statistics, neural network algorithm, obtain the power of different risks and assumptions
Weight determines the Main Factors for influencing risk according to the size of weight.Specific residential quarters collateral risk grading impact factor is as follows:
Data collection
The residential quarters collateral risk factor obtained according to step 2 collects fingers at different levels by analyzing the intension of each risks and assumptions
Mark data.Data collection is using web crawlers and the method manually combined.
Different algorithm models is built according to different risks and assumptions
By the data collection of step 3, the related data of different stage index is obtained, while being built according to different types of index
Corresponding subalgorithm model-such as K-MEANS, SpectralClustering clustering method, the times such as ARIMA, GM (1,1)
Series model.
Major risk factor: cell real factor algorithm uses clustering；
Comprehensively considering single urban cells periphery influences the risks and assumptions of cell real factor, using unsupervised approaches, such as it is poly-
All sample point cells are polymerized to three classes by alanysis algorithm, so that the sample point similitude within class is as big as possible, and class and class
Between similitude it is as small as possible, then according to the otherness between class and class, different community is divided according to different attribute
Class.Then stochastic simulation is sampled, and manually appraises and decides the physical tags of sampling cell, the indexs such as statistics recall rate, accuracy.This is poly-
Alanysis mainly uses K-Means, spectral clustering model.Operation logic is described below:
Clustering (cluster analysis) is one group of group (clusters) that research object is divided into opposite homogeneity
Statistical analysis technique.The operation logic of clustering is as follows:
K-Means clustering algorithm principle:
K-Means is one of clustering algorithm, and wherein K indicates that classification number, Means indicate mean value.As the term suggests K-Means is
A kind of algorithm that data point is clustered by mean value.K-Means algorithm by preset K value and each classification just
The prothyl heart divides similar data point.And optimal cluster result is obtained by the mean iterative optimization after dividing.
K-means algorithm basic step:
(1) select k object as initial mass center from data;
(2) each clustering object is calculated to the distance of mass center, and each clustering object is divided into nearest mass center, forms K cluster；
(3) mass center for calculating each cluster again, forms new mass center；
(4) step (2) and (3) are repeated, until mass center is no longer changed.
SpectralClustering algorithm principle:
SpectralClustering is the algorithm being evolved from graph theory, is widely used in cluster later.
Its main thought is the point all data regarded as in space, can be connected with side between these points.Distance is farther out
Two points between side right weight values it is lower, and the side right weight values between two points being closer are higher, by all numbers
The figure of strong point composition carries out cutting figure, makes side right between subgraph different after cutting figure heavy and low as far as possible, and the side right weight in subgraph
It is high as far as possible, to achieve the purpose that cluster.
SpectralClustering algorithm steps:
Input: sample set D, the generating mode of similar matrix, the dimension m after dimensionality reduction, clustering method, the dimension k after cluster
Output: cluster divides
(1) the similar matrix S of sample is constructed according to the generating mode of the similar matrix of input;
(2) adjacency matrix W, building degree matrix D are constructed according to similar matrix S;
(3) Laplacian Matrix L, L=D-W are calculated;
(4) Laplacian Matrix after building standardization;
(5) the corresponding feature vector of the smallest m characteristic value institute is calculated;
(6) feature vector is standardized, the eigenmatrix F of final composition dimension;
(7) sample tieed up to every a line in F as one, total n sample are clustered with the clustering method of input, are clustered
Dimension is k;
(8) cluster division is obtained.
Major risk factor: cell short-term market price fluctuation factor prediction.Utilize cell history Time Series of Random Macro-price number
According to settling time serial correlation model finds the trend that the price of the cell changes over time, and prejudges price in this, as basis
In the substantially tendency of following a period of time.Then political affairs are regulated and controled according to the amount of increase and amount of decrease combination brisk market index of anticipation price, each city
Plan etc. is comprehensive to measure the price fluctuation risk of cell future in a short time.It is based primarily upon the price fluctuation risk of ARIMA model prediction.
The operation logic of ARIMA model is as follows:
ARIMA model full name is that autoregression integrates moving average model (Autoregressive Integrated Moving
Average Model is abbreviated ARIMA), also known as box-jenkins model, Bock think of-Jenkins method.Wherein ARIMA(p, d,
Q) it is known as difference ARMA model, AR is autoregression, and p is autoregression item；MA is rolling average, and q is mobile flat
Equal item number, the difference number that d is done when becoming steady by time series.So-called ARIMA model, refers to nonstationary time series
It is converted into stationary time series, then only the present worth to its lagged value and stochastic error and lagged value carry out by dependent variable
Return established model.ARIMA model is according to whether former sequence steady and difference of contained part in returning, including movement
Averaging process (MA), autoregressive process (AR), autoregressive moving-average (ARMA) process (ARMA) and ARIMA process.
The basic ideas of ARIMA model are: will predict object over time and formed data sequence be considered as one with
Machine sequence, with certain mathematical model come this sequence of approximate description.This model can be from time sequence after identified
The past value of column and value predicts future value now.
The basic program of ARIMA model prediction
(1) according to the scatter plot of time series, auto-correlation function and partial autocorrelation function figure with ADF unit root test its variance,
Trend and its Rules of Seasonal Changes identify the stationarity of sequence.The time series of history housing price fluctuation is not flat
Steady sequence.
(2) tranquilization processing is carried out to non-stationary series.If data sequence is non-stable, and there are certain growths
Or downward trend, then it needs to carry out difference processing to data, if data there are Singular variance, need to carry out technical office to data
Reason, until the auto-correlation function value and deviation―related function value of treated data are without significant different from zero.
(3) according to the recognition rule of time series models, corresponding model is established.If the deviation―related function of stationary sequence is
Truncation, and auto-correlation function is hangover, can conclude that sequence is suitble to AR model；If the deviation―related function of stationary sequence is hangover
, and auto-correlation function is truncation, then can conclude that sequence is suitble to MA model；If the deviation―related function and auto-correlation of stationary sequence
Function is hangover, then sequence is suitble to arma modeling.(truncation refers to the auto-correlation function (ACF) of time series or partially from phase
Close the property (such as PACF of AR) that function (PACF) is 0 after certain rank；Hangover is that ACF or PACF are not after certain rank
0 property (such as ACF of AR).)
(4) parameter Estimation is carried out, is checked whether with statistical significance.
(5) hypothesis testing is carried out, whether diagnosis residual sequence is white noise.
(6) forecast analysis is carried out using the model for having passed through inspection.
Major risk factor: the prediction of cell short-term market price fluctuation factor is carried out auxiliary with gray prediction GM (1,1) model
Help operation.
11 principles and methods of Grey Theory GM:
1 sets time seriesThere is n observed value,, pass through the new sequence of Accumulating generation, then GM(1,1) and the corresponding differential equation of model are as follows:
Wherein: α is known as developing grey number；μ is known as the interior grey number of raw control.2, it setsFor parameter vector to be estimated,, can benefit
It is solved with least square method.It solves:
The differential equation is solved, prediction model can be obtained:
The gray prediction, which is examined, generally residual test, degree of association inspection and posterior difference examination.
Major risk factor: locational factor part index number algorithm.Nemerow Index method be for describe to detest in range because
Element is to the venture influence degree of residential quarters in its distance range, and when there is a factor distance especially close in a number of factors
When, value decline is obvious, when these values all farther out when, which will increase.
The basic calculating formula of NeiMeiLuo Index are as follows:
Complete model buildings
After completing each two-level index modeling of cell risk rating, using Delphi method, comprehensive each expert weighs risks and assumptions
After the opinion of weight, after the weight of different risks and assumptions and the weight of first class index are obtained by the principal component analysis in statistics,
Total model of national cell risk rating is determined by the size of weight.Each urban cells risk in the whole nation is sought by model batch
Grading.
Note: 1, since the city of national different energy levels is different to different risks and assumptions weights influences, therefore the model is certain
It is also slightly distinguished in index weights marking.
2, the model be built upon real-estate market environment it is more stable in the case where Risk Evaluation model, when real estate city
Field occurs biggish systematicness and changes, such as: having the policy of regulation and control of larger impact to room rate；Or there is the economy of larger impact to room rate
When risk, model must be adjusted.
Adjust model
The cell risk rating obtained for previous step as a result, using random sampling function, sample point that will randomly select
It is sent to expert's testing model result of different cities.Expert feedback result and model result comparative analysis, while passing through engineering
Practise, adjusting and optimizing model so that result be really more nearly.
The publication of residential quarters collateral risk rating model
After completing assessment to cell collateral risk grading mould, can Issuance model, while carrying out Related product design and displaying.
Those skilled in the art can carry out various remodel and change to the present invention.Therefore, present invention covers fall into
Various in the range of appended claims and its equivalent remodel and change.
Claims (4)
1. the batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk, which is characterized in that packet
Include following steps:
S1: residential quarters collateral risk grading intension is defined；
S2: identification residential quarters collateral risk grading risks and assumptions are based on Real Estate Appraising Theory, using Delphi method and statistics
Related algorithm carries out the judgement of real estate risks and assumptions in, after determining the Main Factors for influencing real estate risk, and then identifies
The major risk factor of residential quarters collateral risk grading out；
S3: data collection, the residential quarters collateral risk factor obtained according to step S2, by analyzing in each risks and assumptions
Contain, collects achievement datas at different levels；
S4: building different algorithm models according to different risks and assumptions, by the data collection of step S3, obtains different stage and refers to
Target related data, while corresponding subalgorithm model is built according to different types of index；
S5: model buildings are completed；
S6: seeking the risk rating of National Residential cell in batches as a result, according to the model that step S5 is completed, to National Residential cell batch
Amount grading；
S7: adjustment model, for step S6 obtain risk rating as a result, providing model by the expert of real estate different field
Assessment result, comparative analysis, adjusting and optimizing model parameter so that result be really more nearly；
S8: the publication of National Residential cell collateral risk rating model issues the measurement model of risk rating, while carrying out product exhibition
Show.
2. the batch grading according to claim 1 based on K-MEANS and ARIMA model prediction residential quarters collateral risk
Method, it is characterized in that: method of data capture is using web crawlers and the method manually combined in the step S3.
3. the batch grading according to claim 1 based on K-MEANS and ARIMA model prediction residential quarters collateral risk
Method, it is characterized in that: the step S4 neutron algorithm model, comprising: cell real factor part index number algorithm is directed to, using poly-
Alanysis, i.e. K-Means, SpectralClustering；For market fluctuation predicted portions index algorithm, using ARIMA, GM
(1,1)；For locational factor part index number algorithm, using Nemerow Index method.
4. the batch grading according to claim 1 based on K-MEANS and ARIMA model prediction residential quarters collateral risk
Method is realized it is characterized in that: the step S5 completes model buildings especially by following sub-step:
S51 obtains the score value of each risks and assumptions in two-level index by the calculating of all kinds of submodels of S4；
S52 determines Raw performance weight by algorithm with regress analysis method, in conjunction with Delphi expert graded, determines that every grade of risk refers to
Mark the final score value of weight；
S53 calculates the total score of each cell according to determining index score value and weighting weight；
S54 analyzes the Statistical Distribution of indexs at different levels, and combines clustering algorithm, evaluates the risk class of each cell.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811249353.XA CN109214598A (en) | 2018-10-25 | 2018-10-25 | Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811249353.XA CN109214598A (en) | 2018-10-25 | 2018-10-25 | Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109214598A true CN109214598A (en) | 2019-01-15 |
Family
ID=64996321
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811249353.XA Pending CN109214598A (en) | 2018-10-25 | 2018-10-25 | Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109214598A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738565A (en) * | 2019-10-11 | 2020-01-31 | 中山市银鹿金科信息科技有限公司 | Real estate finance artificial intelligence composite wind control model based on data set |
| CN112488805A (en) * | 2020-12-17 | 2021-03-12 | 四川长虹电器股份有限公司 | Long-renting market early warning method based on multiple regression time series analysis |
-
2018
- 2018-10-25 CN CN201811249353.XA patent/CN109214598A/en active Pending
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738565A (en) * | 2019-10-11 | 2020-01-31 | 中山市银鹿金科信息科技有限公司 | Real estate finance artificial intelligence composite wind control model based on data set |
| CN112488805A (en) * | 2020-12-17 | 2021-03-12 | 四川长虹电器股份有限公司 | Long-renting market early warning method based on multiple regression time series analysis |
| CN112488805B (en) * | 2020-12-17 | 2022-03-25 | 四川长虹电器股份有限公司 | Long-renting market early warning method based on multiple regression time series analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | Risk prediction in financial management of listed companies based on optimized BP neural network under digital economy | |
| Wang et al. | Data-driven mode identification and unsupervised fault detection for nonlinear multimode processes | |
| CN109409677A (en) | Enterprise Credit Risk Evaluation method, apparatus, equipment and storage medium | |
| CN107633265A (en) | For optimizing the data processing method and device of credit evaluation model | |
| CN107194803A (en) | P2P net loan borrower credit risk assessment device | |
| CN106875270A (en) | A kind of method and system design for building and verifying credit scoring equation | |
| CN105022825A (en) | Financial variety price prediction method capable of combining financial news mining and financial historical data | |
| CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
| CN107609771A (en) | A kind of supplier's value assessment method | |
| CN108549907A (en) | A kind of data verification method based on multi-source transfer learning | |
| CN112418476A (en) | Ultra-short-term power load prediction method | |
| CN109685321A (en) | Event risk method for early warning, electronic equipment and medium based on data mining | |
| Ruyu et al. | A comparison of credit rating classification models based on spark-evidence from lending-club | |
| Zekić-Sušac et al. | Cluster analysis and artificial neural networks in predicting energy efficiency of public buildings as a cost-saving approach | |
| CN109214598A (en) | Batch ranking method based on K-MEANS and ARIMA model prediction residential quarters collateral risk | |
| CN111931992A (en) | Power load prediction index selection method and device | |
| CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
| CN114219225A (en) | Power grid investment benefit evaluation system and evaluation method based on multi-source data | |
| CN104143117B (en) | Method for extracting correlation coefficient between special load and daily load of power grid | |
| CN112950048A (en) | National higher education system health evaluation based on fuzzy comprehensive evaluation | |
| Maulana et al. | Crude Oil Price Forecasting Using Long Short-Term Memory | |
| CN112767126A (en) | Collateral grading method and device based on big data | |
| CN117132383A (en) | Credit data processing method, device, equipment and readable storage medium | |
| Mao et al. | Naive Bayesian algorithm classification model with local attribute weighted based on KNN | |
| CN114372694A (en) | College student behavior analysis system |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190115 |