CN104751007B - Computational methods based on building value assessment and device - Google Patents
Computational methods based on building value assessment and device Download PDFInfo
- Publication number
- CN104751007B CN104751007B CN201510181198.2A CN201510181198A CN104751007B CN 104751007 B CN104751007 B CN 104751007B CN 201510181198 A CN201510181198 A CN 201510181198A CN 104751007 B CN104751007 B CN 104751007B
- Authority
- CN
- China
- Prior art keywords
- building
- value
- function
- determined
- assessed
- 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.)
- Active
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
This application discloses a kind of computational methods and device based on building value assessment.One specific implementation mode of the method includes:Obtain the configuration data of multiple buildings;Based on configuration data, function distributed model is established to multiple buildings using machine learning method, obtains the functional character distribution of multiple buildings;The similarity of the building and building to be assessed that have determined that value is calculated according to functional character distribution;It determines at least one and maximum building for having determined that value of building similarity to be assessed, constitutes similar building set;And the value of building to be assessed is determined based on the value of owned building in similar building set.The embodiment can effectively utilize the different configuration attribute features of building, and higher accuracy and better interpretation can be obtained when carrying out value assessment to building.
Description
Technical field
This application involves technical field of data processing, and in particular to for the technical field of data processing of assessment, especially relates to
And the computational methods based on building value assessment and device.
Background technology
In the prior art, the method for the assessment of building value is mainly based upon various configurations and the macroscopic view of building
Market index combination linear regression analysis is predicted.A kind of common prediction technique is to use enjoyment price model
(Hedonic Pricing Model), input model after the configuration feature (such as area, geographical location etc.) of building is quantified
The middle price for calculating building.Wherein by the configuration feature of building be converted to numerical value carry out regression analysis it is relatively difficult, not just
When conversion may result in lower precision of prediction.In addition, there are some configuration features to connect each other, such as total price,
Per square meter price and the gross area, the interpretation of model is poor in the above-mentioned methods, and it is special can not intuitively to embody these configurations
Association between sign.When being calculated using model, it is possible that the positive negative of coefficient between these inter-related configuration features
It is supporting as a result, to influence the precision of value estimations.
Invention content
To solve above-mentioned defect in the prior art, it is intended to provide a kind of configuration feature pair can make full use of building
The method that building carries out value assessment, further, also it is desirable to which a kind of building price of the configuration feature based on integration is provided
It is worth appraisal procedure, to promote the accuracy of assessment.This application provides computational methods and device based on building value assessment.
On the one hand, this application provides a kind of computational methods based on building value assessment.This method includes:It obtains more
The configuration data of a building, plurality of building include building to be assessed and at least one building for having determined that value
Object;Based on configuration data, function distributed model is established to multiple buildings using machine learning method, obtains multiple buildings
Functional character is distributed;The similarity of the building and building to be assessed that have determined that value is calculated according to functional character distribution;Really
The fixed at least one and maximum building for having determined that value of building similarity to be assessed, constitutes similar building set;With
And the value of building to be assessed is determined based on the value of owned building in similar building set.
On the other hand, this application provides a kind of computing devices based on building value assessment.The device includes:It obtains
Unit, the configuration data for obtaining multiple buildings, plurality of building include building to be assessed and at least one
Have determined that the building of value;Modeling unit establishes multiple buildings using machine learning method for being based on configuration data
Function distributed model obtains the functional character distribution of multiple buildings;Computing unit, for being calculated according to functional character distribution
Determine the similarity of the building and building to be assessed of value;Determination unit, for determining at least one and building to be assessed
The maximum building for having determined that value of object similarity, constitutes similar building set;And assessment unit, for based on similar
The value of owned building determines the value of building to be assessed in building set.
Computational methods and device provided by the present application based on building value assessment, pass through matching based on multiple buildings
It sets data and function distributed model is established using the method for machine learning, incorporate the configuration data of building, configuration data is turned
It is changed to functional character distribution, and then similar building set is determined according to functional character distribution, and based on the valence of similar building
Value makes assessment to the value of building to be assessed, efficiently utilizes the heterogeneous characteristic of building, can obtain higher standard
True property and better interpretation.
Description of the drawings
Non-limiting embodiment is described in detail with reference to made by the following drawings by reading, other features,
Objects and advantages will become more apparent upon:
Fig. 1 shows the attribute data of building and the relation schematic diagram of function;
Fig. 2 shows according to the schematical of the computational methods based on building value assessment of the application one embodiment
Flow chart;
Fig. 3 shows the schematical flow of the method for establishing function distributed model according to the application one embodiment
Figure;
Fig. 4 is shown according to the schematic of the computational methods based on building value assessment of the application another embodiment
Flow chart;
Fig. 5 shows the schematic knot of the computing device based on building value assessment according to the application one embodiment
Composition;
Fig. 6 shows the structure of the computer system suitable for the terminal device or server that are used for realizing the embodiment of the present application
Schematic diagram.
Specific implementation mode
The application is described in further detail with reference to the accompanying drawings and examples.It is understood that this place is retouched
The specific embodiment stated is used only for explaining related invention, rather than the restriction to the invention.It also should be noted that in order to
Convenient for description, is illustrated only in attached drawing and invent relevant part with related.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.The application is described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.
Referring to FIG. 1, it illustrates the relation schematic diagrams of the attribute data of building and function.In embodiments herein
In, the configuration data of building is associated with the function that building may have.As shown in Figure 1, building 100 may have
Multiple functions:Function 1, function 2 ... function M.Building 100 can also include multiple configuration datas:Configuration 1, configuration 2, configuration
3 ... N is configured, wherein M, N is positive integer.Each function of building can be associated with each configuration data, each to configure
Data can potentially correspond to multiple functions.Specifically incidence relation can be obtained by statistical data, and can be by proportional numbers
According to indicating.In some implementations, show that each is matched after the configuration data and function of the building of big data quantity can be counted
Set the correspondence between data and its function.For example, in configuration data area be more than 200 square metres of building quantity account for it is all
The 5% of the building total quantity of residential function, area are that 100-200 square metres of building quantity accounts for building for all residential functions
The 40% of object total quantity is built, then area in the configuration data of building can be more than 200 square metres and area is put down in 100-200
Incidence relation between square rice between residential function is quantified as 0.05 and 0.4 respectively.It is established respectively in this way, statistical data can be based on
Incidence relation between configuration data and each function.
Referring to FIG. 2, it illustrates the computational methods based on building value assessment according to the application one embodiment
Schematical flow chart.The present embodiment is mainly applied to server or terminal device with data-handling capacity in this way
In illustrate.
As shown in Fig. 2, in step 201, obtaining the configuration data of multiple buildings.
In the present embodiment, server or terminal device can obtain the configuration data of multiple buildings first.It is wherein more
A building may include building to be assessed and at least one building for having determined that value.Have determined that the building of value
The acquisition modes of configuration data include but not limited to:Collect building data on network, from including building configuration data
Manual research and the building data of label are extracted and obtained in advertisement.The value of these buildings can be by price, appreciation
The data such as potentiality, expected investment repayment ratio intuitively characterize.
In some implementations, the configuration data of building at least may include:Structured data, environmental data and geographical position
Set data.Wherein structured data may include:The data such as area, height, layout, finishing, such as the structured data in house can be with
Including floor, the house type in house, the finishing situation in house, the daylighting situation in house and parking stall number where floor space, house
The data such as amount.Environmental data can characterize the surrounding enviroment for influencing building value, may include periphery natural environment, traffic shape
The data such as condition, medical facilities, service facilities.Such as may include the air quality on periphery, sound level, water quality, greening
The data such as area, can also include Landscape Region, park, hospital, supermarket, Catering Pubs, school, public traffic station, residential block,
Range data between the quantity in the places such as market, bank, office building, recreation ground, factory, type and each place and building.Ground
Reason position data may include the actual geographic position data of building and opposite geographic position data, such as the geographical position in house
It may include that the location of house (such as certain street number) and the house are opposite in a big region to set data
Position (such as house is located at the North 4th Ring Road of Beijing).
In step 202, it is based on configuration data, function distributed mode is established to multiple buildings using machine learning method
Type obtains the functional character distribution of multiple buildings.
In general, each building has multiple potential functions.Such as house can have school district room, house
The functions such as room, investment room, upper jail.These functions are accomplished by the different configuration datas of building, i.e., these functions with
There are relevances as shown in Figure 1 for the configuration data of building.Therefore, building can be expressed as it is several functionally
Distribution, and then the value of building is assessed according to function distribution.
In the present embodiment, modeling analysis can be carried out to the configuration data of building.Machine learning method may be used
The function distributed model for establishing building obtains the function distributed model of building.It in some implementations, can be based on acquisition
The configuration data of multiple buildings establishes function distributed model using topic model.It can be using building as in topic model
Document, the function of building is as the theme in topic model, and each configuration data is as the word in subject document, statistics
The configuration data of multiple buildings calculates the conditional probability that each configuration data corresponds to each function.With the configuration number in house
According to be within 1 km of periphery public transport station points be more than 5, function be working function for, can count 1 km of periphery by
Public transport station points account for the ratio of all statistics buildings more than 5 building, and obtain working function by machine learning
It goes to work work(in the ratio of building of the public transport station points more than 5 and the function of each building within 1 km of middle periphery
The probability of energy.In machine-learning process, the probability for multiple functions that building may have can be sequentially or simultaneously calculated, then
It can obtain the functional character distribution of each building.
In above-mentioned machine learning method, when having relevance between each configuration data, design conditions probability can be passed through
These associated configuration datas are integrated, such as by Landscape Region incremental data and green coverage Data Integration are the greening number of degrees
According to, and the greening degrees of data based on integration calculates the probability of each function of building.Function distributed model uses during establishing
The conditional probability of configuration data and function can not only reflect influence size of each configuration data to each function, but also with more
Good interpretation.
In some implementations, each function is not specific in the functional character distribution by modeling obtained building
At this moment meaning can explain each function in conjunction with domain knowledge.Such as function may include investment function, school district work(
Energy, quotient area function, inhabitation function, industrial function etc. so that function distributed model has better interpretation.
Functional character by the way that the configuration data of building to be converted to numerical value expression is distributed, by configuration data and function phase
Association, can effectively utilize the isomerism of building, you can to efficiently use the different configuration features of building, and then ensure
The estimation precision of building value.
In step 203, the phase of the building and building to be assessed that have determined that value is calculated according to functional character distribution
Like degree.
The functional character distribution that model in based on step 202 calculates the building for having determined that value is built with to be assessed
After building the functional character distribution of object, can be distributed according to functional character calculate multiple buildings for having determined that value with it is to be assessed
The similarity of building.As described above, may include corresponding at least one function and each function in functional character distribution
Probability.Functional character distribution can be expressed as vector, and each element can indicate that a kind of function and function institute are right in vector
The probability answered.Calculating formula of similarity can be utilized to calculate the function or work(of the building and building to be assessed that have determined that value
The similarity that can be distributed, the similarity as the building and building to be assessed for having determined that value.The similarity can characterize
Have determined that the similarity degree or degree of correlation between the building of value and building to be assessed.Similarity is bigger, then can recognize
It is higher for similarity degree or correlation.Optional similarity calculating method may include cosine similarity calculating, pearson correlation
Coefficient calculating etc..
In some embodiments, the building for having determined that value and building to be assessed are calculated according to functional character distribution
Similarity may be used under type such as and carry out:It calculates and has determined that the building of value is distributed with the functional character of building to be assessed
In probability corresponding to each function similarity, obtain function phase of the building with building to be assessed for having determined that value
It is distributed like degree, determines the similarity of the building and building to be assessed that have determined that value based on the distribution of function similarity later.
Specifically, the inverse of the absolute difference between the probability corresponding to each function can be calculated, and (i.e. maximum phase is normalized
It is 1), as the corresponding similarity value of the function like degree.If in the functional character distribution of a certain building for having determined that value
Including the function being not present in the functional character distribution of building to be assessed, it may be considered that the function is in building to be assessed
Corresponding probability is zero in functional character distribution.In this way, to the building of each known value, each function can be obtained
Corresponding similarity, to which the function similarity of the building and building to be assessed that obtain each known value is distributed.
When the functional character of building, which is distributed, to be indicated with vector, the distribution of function similarity can also be indicated with vector, each in vector
The value of a element can be the value of function similarity.
It is possible to further any one following feature for being distributed function similarity as have determined that the building of value with
The similarity of the building to be assessed:Weighted sum, weighted average, mean value, mean square deviation and standard deviation.When function similarity
When distribution is indicated with vector, each element in vector represents a function, and the value of the element represents the work(corresponding to the function
The size of energy similarity.The weighted sum of all elements value, weighted average, mean value, mean square deviation or mark in the vector can be calculated
It is accurate poor, obtain a numerical value, the similarity as the building and the building to be assessed for having determined that value.
In the present embodiment, the building of value is had determined that each, is calculated it in the manner described above and is built with to be assessed
The similarity of object is built, then can obtain the similarity of each building for having determined that value and building to be assessed.
In step 204, at least one and maximum building for having determined that value of building similarity to be assessed is determined,
Constitute similar building set.
After obtaining the similarity of multiple buildings for having determined that value with building to be assessed, can to similarity into
Row sequence, chooses at least one and maximum building of building similarity to be assessed from all buildings for having determined that value
Object constitutes similar building set.By step 203 calculated similarity be the similarity being distributed based on functional character,
Thus the building in similar building set may be considered building similar with building functions to be assessed.
In step 205, the value based on owned building in similar building set determines the valence of building to be assessed
Value.
In the present embodiment, after determining similar building set, can be estimated according to the value of building in the set
Calculate the value of building to be assessed.Can include multiple buildings similar with building functions to be assessed in similar building set
Object.It in some implementations, can be to multiple and building to be assessed to make the value assessment accuracy higher of building to be assessed
The value of intimate building is averaged, using the average value of the value of similar building as the value of building to be assessed
Assessment result.
In other realizations, summation can also be weighted to the value of owned building in similar building set,
With the value of determination building to be assessed.Under normal circumstances, the value assessment of building and assessment time correlation.Similar building
Have determined that the value of the building of value determines that the time may be with the value assessment association in time journey of building to be assessed in set
Degree is different.The time is assessed with similar in the value determining time, temporal associativity is stronger, and assesses the time and determine time phase with value
Away from farther out, temporal associativity is poor.At the same time it can also consider Different Effects of the similarity size to value assessment.Therefore, may be used
Weight is arranged for the building in similar building set based on above-mentioned consider.The weight can be based on building to be assessed with
Similarity and/or temporal associativity between the building of value are had determined that in similar building set to determine.If with γ tables
Show time decay factor, siThe similarity of the building and building to be assessed that have determined that value for i-th, n indicate to determine valence
Time span between time of value and assessment time has determined that the weight w of the building of value in similar building setiIt can
To be calculated according to formula (1):
wi=si*(1-γ*Sigmoid(n)) (1)
Wherein, Sigmoid (n) is the Sigmoid functions using n as variable.It can not consider that the time decays as γ=0,
wi=si.As γ ≠ 0, weight wiLength n growths at any time are gradually reduced.
If the value of i-th of building has been confirmed as P in similar building seti, then the valence of building to be assessed
Value PP can be calculated according to formula (2):
Weighted average is done by the value to multiple buildings, according to the phase between different buildings and building to be assessed
Building value proportion shared in building value assessment to be assessed, Ke Yiti are determined like degree and temporal associativity size
Rise reliability and accuracy that building value assessment calculates.
For above-described embodiment of the application, application scenarios can be:Match confidence based on the current source of houses and the history source of houses
The room rate of breath and the history source of houses estimates the price of the current source of houses.Assuming that each source of houses there are potential various functions, such as learn
Area room function, investment function, residential function, upper jail function etc..These functions can pass through the different configuration informations of the source of houses
Come accomplished.Machine learning method may be used to model the source of houses, obtain the distribution of the different sources of houses functionally, later
It can be distributed the similarity calculated between the current source of houses and the history source of houses according to function, it is similar to find function using similarity
The history source of houses, estimate the price of the current source of houses using the price of these intimate history sources of houses.In estimation procedure,
It is contemplated that the similarity size of the history source of houses and the current source of houses and influence of the time difference to room rate is sold, for the history source of houses
Weight is arranged in room rate, is weighted averagely to the price of multiple similar sources of houses, obtains the prediction price to the current source of houses.It is optional
Ground can also utilize prediction price subtract current bid price, then divided by current bid price, obtain the upside potential of the current source of houses.
The above embodiments of the present application can be applied not only to house price Prediction System in above application scene and
House upside potential assessment system can be also used in abnormal House to let detecting system and house commending system.For example, different
It often in House to let detecting system, can calculate the value in house according to the method for above-described embodiment, and actually be sold with house
Valence compares, if the two difference is larger, it is believed that the house is the house sold extremely.In another example in house commending system
In, the similarity in house can be calculated according to the method for above-described embodiment, and be based on similarity size to user's recommendation function phase
As house;The estimated price that multiple houses to be selected can also be calculated according to the method for above-described embodiment, based on house to be selected
Estimated price recommends the house for meeting user's expected price;Can also based on house prediction price and actual selling price calculate
After the upside potential in house, recommend the house of high upside potential to user.
The method that the above embodiments of the present application provide passes through based on the building and building to be assessed for having determined that value
Configuration data, using machine learning method obtain functional character be distributed, calculate similarity later, determine similar building set,
And according to the value of the value calculation of similar building building to be assessed.The heterogeneous characteristic of building can be effectively utilized,
And the reliability and accuracy of building value assessment can be promoted by configuration data of the integration with relevance.
With further reference to Fig. 3, it illustrates according to the method for establishing function distributed model of the application one embodiment
Schematical flow chart, namely show a kind of flow chart of realization method of step 202 in embodiment illustrated in fig. 2.
As shown in figure 3, in step 301, more than one that multiple buildings may have the function of is determined.
In the present embodiment, the statistical data that can rule of thumb or by the information searches mode such as network obtain first is true
Determine more than one that building may have the function of.For example, by disclosed building functions information on network or can pass through
The function of having determined that the building of value has is known in the building training of big data quantity, such as school district function, quotient area function, lives
The functions such as function, investment function, industrial function, then assume that building to be assessed also has the function of these.
In step 302, configuration data is associated with each possible function, initialize building functions distributed mode
Type.
As described above, each function of building can be associated with each configuration data, each configuration data can dive
Multiple functions are corresponded on ground.Each configuration data can potentially correspond to multiple functions.In some implementations, configuration data and work(
The incidence relation of energy can be obtained by statistical data, and can be indicated by ratio data or weight.In the present embodiment, it builds
The function distributed model for building object can include this incidence relation, such as can be a pass with each configuration data and each function
Contact number is input, is the probability Distribution Model exported with the probability of function and function.In the function distribution of initialization building
When model, ratio data that can be based on above-mentioned statistics or weight setting configuration data are initial with the incidence coefficient of each function
Value, can also be randomly provided the initial value of incidence coefficient, obtain the function distributed model of initialization.
In step 303, it using the configuration data of multiple buildings as training sample, is built using machine learning method estimation
Build the parameter of object function distributed model.
In the present embodiment, sample set can be trained by using machine learning method to adjust function distributed mode
The parameter of type obtains reliable and stable incidence coefficient value and function distributed model.Wherein sample set can be above-mentioned multiple builds
The configuration data of object is built, the machine learning method of use may be, for example, matrix decomposition, topic model etc. based on limitation.At some
, can also be according to the function for the building for having determined that value come the error of computation model in realization, and it is based on error transfer factor model
Parameter.Such as when it is quotient area function to have determined that the function of a building of value, if calculated in the training process
The probability very little of the quotient area function of the building, then can will configuration data (such as periphery eating and drinking establishment associated with quotient area function
Spread quantity, residential block quantity, bank quantity) incidence coefficient be turned up;Or multiple buildings can be counted by function distributed mode
Difference between the function and actual functional capability of the maximum probability that type determines, as the error of current function distributed model, if accidentally
Difference is more than threshold value, then constantly can carry out correction adjustment to the parameter of model, (be, for example, less than threshold when the error of model is sufficiently small
Value) when, it is believed that current function distributed model parameter is suitable parameter.
In step 304, the parameter based on building functions distributed model calculates the functional character distribution of multiple buildings.
After the parameter for determining function distributed model, it may be determined that function distributed model.The model can be by configuration data
Be converted to functional character distribution.The configuration data of the building for having determined that value and building to be assessed is inputted the model then may be used
To calculate the probability corresponding to the potential function of these buildings and each function, you can to obtain the work(of multiple buildings
It can feature distribution.
Referring to FIG. 4, it illustrates the calculating sides based on building value assessment according to another embodiment of the application
The schematical flow chart of method.
As shown in figure 4, in step 401, obtaining the configuration data of multiple buildings.
Server or terminal device can obtain the configuration data of multiple buildings first.Plurality of building can wrap
Include building to be assessed and at least one building for having determined that value.The configuration data of building can obtain in the following way
It takes:The building data on network are collected, extracts from the advertisement comprising building configuration data and obtains manual research simultaneously
The building data of label.
In some implementations, the configuration data of building at least may include:Structured data, environmental data and geographical position
Set data.Wherein structured data may include:The data such as area, height, layout, finishing.Environmental data can characterize influence and build
The surrounding enviroment data for building price value may include periphery natural environment, traffic, medical facilities, service facilities etc.
Data.Geographic position data may include the actual geographic position data of building and opposite geographic position data.
In step 402, extensive processing is carried out to configuration data.
May include some non-quantized data in the configuration data of building, such as finishing situation, natural environment, geography
The data such as position.These data do not have specific numerical value usually and indicate.In the present embodiment, these data can be carried out general
Change is handled.It in some implementations, can be to by continuous configuration data discretization, being compiled to the configuration data after discretization
Code, the numerical value for obtaining configuration data indicate.Such as configuration data can be quantified, obtain the configuration number indicated by discrete data
According to then discrete data can be encoded to 0,1 feature, a discrete data with T value (T is positive integer) is made to convert
The feature for being 0 or 1 for T value.In other realizations, configuration data can be converted to vector characteristics expression.Such as it can
With by natural environment data be converted to comprising water grade, air quality grade, sound level, green coverage grade it is one-dimensional to
Amount.The data that wherein water grade, air quality grade, sound level, green coverage grade can be provided by environmental protection administration obtain
Go out, grade can also be carried out according to specific water quality data, air quality data, noise data, green coverage data by user and commented
It is fixed to determine.
Building generally comprises a variety of different types of configuration datas, in some embodiments, can be to all configuration numbers
According to extensive processing is all carried out, all configuration datas are converted into same data mode and are indicated, such as vector, matrix or ratio
Number of cases value is made each to be trained to model based on same data mode when being modeled based on transformed configuration data
Definitely, the functional character distributed model that training obtains, which has, preferably may be used for the influence that configuration data is distributed building functions
It is explanatory.
In step 403, it is based on configuration data, function distributed mode is established to multiple buildings using machine learning method
Type obtains the functional character distribution of multiple buildings.
In the present embodiment, machine learning method may be used, modeling analysis is carried out to the configuration data of building, obtain
The function distributed model of building.In machine-learning process, can sequentially or simultaneously calculate building may have it is multiple
The probability of function obtains the functional character distribution of each building.
In step 404, the time response based on configuration data and/or statistical property to have determined that the building of value into
Row filtering.
In the present embodiment, in order to reduce the complexity of similarity calculation, time restriction can be increased, filter configuration data
Outmoded building.Specifically, can by the configuration data of acquisition with the built time of building to be assessed or assessment the time
Data apart from each other are rejected.Such as it can be by data calculation time in configuration data and building the built time to be assessed apart
Building corresponding to data or the built time more than 10 years are with building the built time to be assessed at a distance of building more than 20 years
Object is built to filter out.It is also based on the statistical property of configuration data, using the number such as such as k-d tree (i.e. k Wei Shu, k are positive integer)
Configuration data is filtered according to structure, to be filtered to corresponding building.
In step 405, the phase of the building and building to be assessed that have determined that value is calculated according to functional character distribution
Like degree.
The functional character distribution that model in based on step 403 calculates the building for having determined that value is built with to be assessed
After building the functional character distribution of object, can be distributed according to functional character calculate multiple buildings for having determined that value with it is to be assessed
The similarity of building.Functional character distribution can be expressed as vector, and each element can indicate a kind of function and should in vector
Probability corresponding to function.Calculating formula of similarity can be utilized to calculate the building for having determined that value and building to be assessed
Function or the similarity of function distribution, the similarity as the building and building to be assessed for having determined that value.Optional phase
May include cosine similarity calculates, Pearson correlation coefficients calculate etc. like degree computational methods.
In the present embodiment, the building of value is had determined that each, is calculated it in the manner described above and is built with to be assessed
The similarity of object is built, then can obtain the similarity of each building for having determined that value and building to be assessed.
In a step 406, at least one and maximum building for having determined that value of building similarity to be assessed is determined,
Constitute similar building set.
After obtaining the similarity of multiple buildings for having determined that value with building to be assessed, can from it is all really
At least one and maximum building of building similarity to be assessed is chosen in the building of price value, constitutes similar building collection
It closes.
In step 407, the value based on owned building in similar building set determines the valence of building to be assessed
Value.
In the present embodiment, after determining similar building set, can be estimated according to the value of building in the set
Calculate the value of building to be assessed.Can include multiple buildings similar with building functions to be assessed in similar building set
Object.It in some implementations, can be to multiple and building to be assessed to make the value assessment accuracy higher of building to be assessed
The value of intimate building is averaged, and the average value as building to be assessed of the value of similar building is commented
Estimate result.
In other realizations, temporal associativity and similarity are also based on to all buildings in similar building set
The value of object is weighted summation, with the value of determination building to be assessed.
Above-mentioned steps 401,403,405,406 and 407 respectively in previous embodiment step 201,202,203,204 and
205 is identical, and the above-mentioned description for each step in Fig. 2 is also suitable correspondence step in this present embodiment, and details are not described herein again.From
In Fig. 4 as can be seen that unlike embodiment described in Fig. 2, the method that the present embodiment is provided increases the step of extensive processing
Rapid 402 and filtration step 404.It can convert configuration data to numerical characteristics by the step 402 of increase, in order to carry out work(
The calculating and estimation of energy distributed model.The accuracy of similarity calculation can be further promoted by the step 404 of increase, in turn
More accurately determine similar building, and the value of the value estimate building to be assessed based on similar building.
Referring to FIG. 5, it illustrates the computing devices based on building value assessment according to the application one embodiment
500 schematic diagram.As shown in figure 5, the computing device 500 based on building value assessment may include acquiring unit
501, modeling unit 502, computing unit 503, determination unit 504 and assessment unit 505.Acquiring unit 501 can be used for obtaining
Take the configuration data of multiple buildings, plurality of building includes building to be assessed and at least one has determined that value
Building.The configuration data for multiple buildings that modeling unit 502 can be used for obtaining based on acquiring unit 501, using machine
Learning method establishes function distributed model to multiple buildings, obtains the functional character distribution of multiple buildings.Computing unit 503
The functional character distribution that can be used for being obtained according to modeling unit 502 calculates the building for having determined that value and building to be assessed
The similarity of object.Determination unit 504 can be used for determining based on the similarity calculation result of computing unit 503 it is at least one with wait for
The maximum building for having determined that value of building similarity is assessed, similar building set is constituted.Assessment unit 505 can be used
The value of owned building determines the value of building to be assessed in the similar building set determined based on determination unit.
In some implementations, computing unit 503 may include:Function similarity is distributed determination unit and similarity calculation list
Member.Wherein function similarity distribution determination unit can be used for calculating the work(of the building and building to be assessed that have determined that value
The similarity between probability in energy feature distribution corresponding to each function, obtains having determined that the building of value is built with to be assessed
Build the function similarity distribution of object.The work(that similarity calculated can be used for obtaining based on function similarity distribution determination unit
Energy similarity distribution determines the similarity of the building and building to be assessed that have determined that value.
In some embodiments, the above-mentioned computing device 500 based on building value assessment can also include that pretreatment is single
Member 506 and filter element 507.Wherein pretreatment unit 506 can be used for carrying out extensive processing to configuration data.Such as it will be continuous
Configuration data discretization, the configuration data after discretization is encoded, obtain configuration data numerical value indicate;Or it will match
It sets data and is converted to vector characteristics expression.Filter element 507 can be based on configuration data time response and/or statistical property pair
Have determined that the building of value is filtered.
All units described in device 500 are corresponding with reference to each step in figure 1 and the method for Fig. 4 descriptions.As a result,
Operation and feature above with respect to the computational methods description based on building value assessment are equally applicable to device 500 and wherein wrap
The unit contained, details are not described herein.Corresponding units in device 500 can be with the unit phase in terminal device and/or server
Mutually cooperation is to realize the scheme of the embodiment of the present application.
The computing device based on building value assessment that the above embodiments of the present application provide will be configured by modeling unit
Data standard is changed to functional character distribution, and the different of building can be effectively utilized when assessing building value and are configured
Attributive character, and can obtain higher accuracy by configuration attribute feature of the integration with relevance and preferably may be used
It is explanatory.
Fig. 6 shows the structure of the computer system suitable for the terminal device or server that are used for realizing the embodiment of the present application
Schematic diagram.
As shown in fig. 6, computer system 600 includes central processing unit (CPU) 601, it can be read-only according to being stored in
Program in memory (ROM) 602 or be loaded into the program in random access storage device (RAM) 603 from storage section 608 and
Execute various actions appropriate and processing.In RAM 603, also it is stored with system 600 and operates required various programs and data.
CPU 601, ROM 602 and RAM 603 are connected with each other by bus 604.Input/output (I/O) interface 605 is also connected to always
Line 604.
It is connected to I/O interfaces 605 with lower component:Importation 606 including keyboard, mouse etc.;It is penetrated including such as cathode
The output par, c 607 of spool (CRT), liquid crystal display (LCD) etc. and loud speaker etc.;Storage section 608 including hard disk etc.;
And the communications portion 609 of the network interface card including LAN card, modem etc..Communications portion 609 via such as because
The network of spy's net executes communication process.Driver 610 is also according to needing to be connected to I/O interfaces 605.Detachable media 611, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on driver 610, as needed in order to be read from thereon
Computer program be mounted into storage section 608 as needed.
As on the other hand, present invention also provides a kind of computer readable storage medium, the computer-readable storage mediums
Matter can be computer readable storage medium included in device described in above-described embodiment;Can also be individualism, not
The computer readable storage medium being fitted into terminal device.The computer-readable recording medium storage there are one or one with
Upper program, the program can include the program code for method shown in execution flow chart.
Flow chart in attached drawing and block diagram, it is illustrated that according to the system of various embodiments of the invention, device, method and calculating
The architecture, function and operation in the cards of machine program product.In this regard, each box in flowchart or block diagram can
To represent a part for a module, program segment, or code, the part of the module, program segment, or code include one or
Multiple executable instructions for implementing the specified logical function.It should also be noted that in some implementations as replacements, box
Middle marked function can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated
It can essentially be basically executed in parallel, they can also be executed in the opposite order sometimes, this is depended on the functions involved.
It is also noted that the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart, it can
It is realized with the dedicated hardware based systems of the functions or operations as defined in execution, or specialized hardware can be used and calculated
The combination of machine instruction is realized.
Above description is only the preferred embodiment of the application and the explanation to institute's application technology principle.People in the art
Member should be appreciated that invention scope involved in the application, however it is not limited to technology made of the specific combination of above-mentioned technical characteristic
Scheme, while should also cover in the case where not departing from the inventive concept, it is carried out by above-mentioned technical characteristic or its equivalent feature
Other technical solutions of arbitrary combination and formation.Such as features described above has similar work(with (but not limited to) disclosed herein
Can technical characteristic replaced mutually and the technical solution that is formed.
Claims (16)
1. a kind of computational methods based on building value assessment, which is characterized in that the method includes:
Obtain the configuration data of multiple buildings, the multiple building includes building to be assessed and at least one has determined that
The building of value;
Based on the configuration data, function distributed model is established to the multiple building using machine learning method, obtains institute
The functional character distribution of multiple buildings is stated, the functional character distribution includes corresponding at least one function and each function
Probability;
Each function institute in the building of value and the functional character distribution of the building to be assessed is had determined that described in calculating
The similarity of corresponding probability obtains the function similarity point of the building for having determined that value and the building to be assessed
Cloth;
It is distributed based on the function similarity described in determining and has determined that the building of value is similar to the building to be assessed
Degree;
It determines at least one and maximum building for having determined that value of building similarity to be assessed, constitutes similar building
Object set;And
The value of the building to be assessed is determined based on the value of owned building in the similar building set.
2. according to the method described in claim 1, it is characterized in that, described be based on the configuration data, using machine learning side
Method establishes function distributed model to the multiple building, obtains the functional character distribution of the multiple building, including:
Determine more than one that the multiple building may have the function of;
The configuration data is associated with each possible function, initialize building functions distributed model;
Using the configuration data of the multiple building as training sample, the building functions are estimated using machine learning method
The parameter of distributed model;
Parameter based on the building functions distributed model calculates the functional character distribution of the multiple building.
3. according to the method described in claim 1, it is characterized in that, described be distributed described in determination based on the function similarity
Determine the similarity of the building and the building to be assessed of value, including:
Using any one following feature that the function similarity is distributed had determined that as described in the building of value with it is described to be evaluated
Estimate the similarity of building:Weighted sum, weighted average, mean value, mean square deviation and standard deviation.
4. according to the method described in claim 1, it is characterized in that, further including:
Extensive processing is carried out to the configuration data, including:
By continuous configuration data discretization, the configuration data after discretization is encoded, obtains the numerical tabular of configuration data
Show;Or
The configuration data is converted to vector characteristics to indicate.
5. according to the method described in one of claim 1-4, which is characterized in that further include:It is distributed according to the functional character
Before calculating has determined that the building of value and the similarity of the building to be assessed,
Time response and/or statistical property based on the configuration data have determined that the building of value is filtered to described.
6. according to the method described in one of claim 1-4, which is characterized in that described based on institute in the similar building set
There is the value of building to determine the value of the building to be assessed, including:
Based on the building to be assessed to had determined that in the similar building set similarity between the building of value and/
Or temporal associativity, determine the weight for the building that value is had determined that in the similar building set;
Summation is weighted to the value of owned building in the similar building set, with the determination building to be assessed
Value.
7. according to the method described in claim 6, it is characterized in that, being determined in the similar building set according to following formula
Have determined that the weight of the building of value:
wi=si*(1-γ*Sigmoid(n))
Wherein, wiFor had determined that in the similar building set value building weight, γ be time decay factor, si
The similarity of the building and the building to be assessed that have determined that value for i-th, n are the building for having determined that value
Determine the time span between the time of value and assessment time, Sigmoid (n) is the Sigmoid functions using n as variable.
8. according to the method described in one of claim 1-4, the function of the multiple building includes at least:It invests function, learn
Area's function, quotient area function, inhabitation function, industrial function;
The configuration data includes at least:Structured data, environmental data, geographic position data.
9. a kind of computing device based on building value assessment, which is characterized in that described device includes:
Acquiring unit, the configuration data for obtaining multiple buildings, the multiple building include building to be assessed and
At least one building for having determined that value;
Modeling unit establishes function point using machine learning method for being based on the configuration data to the multiple building
Cloth model, obtains the functional character distribution of the multiple building, and functional character distribution includes at least one function and often
Probability corresponding to a function;
Computing unit, for having determined that the building of value to be assessed is built with described described in being calculated according to functional character distribution
Build the similarity of object;
Determination unit, for determining at least one and maximum building for having determined that value of building similarity to be assessed
Object constitutes similar building set;And
Assessment unit, for determining the building to be assessed based on the value of owned building in the similar building set
Value;
Wherein, the computing unit includes:
Function similarity is distributed determination unit, for calculating the building for having determined that value and the building to be assessed
The similarity of probability in functional character distribution corresponding to each function, obtain the building for having determined that value with it is described
The function similarity of building to be assessed is distributed;
Similarity calculated, for have determined that described in being determined based on the function similarity distribution building of value with it is described
The similarity of building to be assessed.
10. device according to claim 9, which is characterized in that the modeling unit is described for obtaining as follows
The functional character of multiple buildings is distributed:
Determine more than one that the multiple building may have the function of;
The configuration data is associated with each possible function, initialize building functions distributed model;
Using the configuration data of the multiple building as training sample, the building functions are estimated using machine learning method
The parameter of distributed model;
Parameter based on the building functions distributed model calculates the functional character distribution of the multiple building.
11. device according to claim 9, which is characterized in that the similarity calculated is by the function similarity
Any one following feature of distribution has determined that the similarity of the building and the building to be assessed of value as described in:Weighting
With weighted average, mean value, mean square deviation and standard deviation.
12. device according to claim 9, which is characterized in that further include:
Pretreatment unit, for carrying out extensive processing to the configuration data;
The pretreatment unit to the configuration data for carrying out extensive processing as follows:
By continuous configuration data discretization, the configuration data after discretization is encoded, obtains the numerical tabular of configuration data
Show;Or
The configuration data is converted to vector characteristics to indicate.
13. according to the device described in one of claim 9-12, which is characterized in that further include:Filter element, for based on described
The time response and/or statistical property of configuration data have determined that the building of value is filtered to described.
14. according to the device described in one of claim 9-12, which is characterized in that the assessment unit is for true as follows
The value of the fixed building to be assessed:
Based on the building to be assessed to had determined that in the similar building set similarity between the building of value and/
Or temporal associativity, determine the weight for the building that value is had determined that in the similar building set;
Summation is weighted to the value of owned building in the similar building set, with the determination building to be assessed
Value.
15. device according to claim 14, which is characterized in that the assessment unit determines the phase according to following formula
Like the weight for the building for having determined that value in building set:
wi=si*(1-γ*Sigmoid(n))
Wherein, wiFor had determined that in the similar building set value building weight, γ be time decay factor, si
The similarity of the building and the building to be assessed that have determined that value for i-th, n are the building for having determined that value
Determine the time span between the time of value and assessment time, Sigmoid (n) is the Sigmoid functions using n as variable.
16. according to the device described in one of claim 9-12, the function of the multiple building includes at least:Investment function,
School district function, quotient area function, inhabitation function, industrial function;
The configuration data includes at least:Structured data, environmental data, geographic position data.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510181198.2A CN104751007B (en) | 2015-04-16 | 2015-04-16 | Computational methods based on building value assessment and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510181198.2A CN104751007B (en) | 2015-04-16 | 2015-04-16 | Computational methods based on building value assessment and device |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104751007A CN104751007A (en) | 2015-07-01 |
CN104751007B true CN104751007B (en) | 2018-07-24 |
Family
ID=53590682
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510181198.2A Active CN104751007B (en) | 2015-04-16 | 2015-04-16 | Computational methods based on building value assessment and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104751007B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8676680B2 (en) | 2006-02-03 | 2014-03-18 | Zillow, Inc. | Automatically determining a current value for a home |
JP6805912B2 (en) * | 2017-03-13 | 2020-12-23 | 横河電機株式会社 | Evaluation device, evaluation system, and evaluation method |
CN107798636A (en) * | 2017-04-11 | 2018-03-13 | 平安科技(深圳)有限公司 | Building information processing method, device, computer equipment and storage medium |
CN107239967A (en) * | 2017-05-10 | 2017-10-10 | 平安科技(深圳)有限公司 | House property information processing method, device, computer equipment and storage medium |
TWI637342B (en) * | 2017-09-08 | 2018-10-01 | 劉劭桓 | Geographical Information Evaluation System for House Prices |
CN109523306A (en) * | 2018-10-23 | 2019-03-26 | 上海舍汇信息技术有限公司 | A kind of multifactor house property valuation system and its estimation method |
CN110619537A (en) * | 2019-06-18 | 2019-12-27 | 北京无限光场科技有限公司 | Method and apparatus for generating information |
CN113742564A (en) * | 2020-05-29 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Target resource pushing method and device |
CN111753362B (en) * | 2020-07-01 | 2021-08-06 | 中安建设安装集团有限公司 | Large building design method based on mobile platform |
CA3192347A1 (en) * | 2020-08-26 | 2022-03-03 | MFTB Holdco, Inc. | Synthetic comparable market analysis systems and methods |
CN112348322A (en) * | 2020-10-23 | 2021-02-09 | 太原理工大学 | Existing residential building performance hierarchical evaluation method |
CN113468627A (en) * | 2020-12-04 | 2021-10-01 | 广东省建筑设计研究院有限公司 | Novel concrete structure building group safety assessment method |
CN113538172B (en) * | 2021-07-12 | 2023-04-18 | 中建西部建设贵州有限公司 | Method and device for intelligently acquiring content of concrete acicular stone |
CN114021948A (en) * | 2021-11-02 | 2022-02-08 | 安徽省招标集团股份有限公司 | Construction cost evaluation method and device based on building similarity |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102254277A (en) * | 2011-06-27 | 2011-11-23 | 中国建设银行股份有限公司 | Data processing system and method for real estate valuation |
CN102831541A (en) * | 2012-08-14 | 2012-12-19 | 上海克而瑞信息技术有限公司 | Automatic assessment method and device of house value |
CN103578057A (en) * | 2012-08-10 | 2014-02-12 | 北京奥齐都市网络科技有限公司 | Real estate value estimation method based on artificial neural network statistic model |
-
2015
- 2015-04-16 CN CN201510181198.2A patent/CN104751007B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102254277A (en) * | 2011-06-27 | 2011-11-23 | 中国建设银行股份有限公司 | Data processing system and method for real estate valuation |
CN103578057A (en) * | 2012-08-10 | 2014-02-12 | 北京奥齐都市网络科技有限公司 | Real estate value estimation method based on artificial neural network statistic model |
CN102831541A (en) * | 2012-08-14 | 2012-12-19 | 上海克而瑞信息技术有限公司 | Automatic assessment method and device of house value |
Non-Patent Citations (3)
Title |
---|
二手房价格评估系统设计与研究;闫敬竹等;《价值工程》;20131231;第168-171页 * |
基于神经网络的房地产估价模型研究及其Matlab实现;张晓平等;《技术经济与管理研究》;20090131;第2009年卷(第1期);第23-25页 * |
基于神经网络的房地产估价研究;王晞;《测绘与空间地理信息》;20131031;第36卷(第10期);第129-132页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104751007A (en) | 2015-07-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104751007B (en) | Computational methods based on building value assessment and device | |
US11238065B1 (en) | Systems and methods for generating and implementing knowledge graphs for knowledge representation and analysis | |
Huang et al. | Geographically and temporally weighted regression for modeling spatio-temporal variation in house prices | |
US20170053297A1 (en) | Enhanced systems, processes, and user interfaces for scoring assets associated with a population of data | |
Anselin et al. | Interpolation of air quality measures in hedonic house price models: spatial aspects | |
US8583562B1 (en) | Predicting real estate and other transactions | |
Morano et al. | Bare ownership evaluation. Hedonic price model vs. artificial neural network | |
US11983203B1 (en) | Systems and methods of establishing correlative relationships between geospatial data features in feature vectors representing property locations | |
Cho et al. | Relationship between value of open space and distance from housing locations within a community | |
Goffette-Nagot et al. | Spatial analysis of residential land prices in Belgium: accessibility, linguistic border, and environmental amenities | |
Palm-Forster et al. | Valuing Lake Erie beaches using value and function transfers | |
US20180330390A1 (en) | Enhanced systems, processes, and user interfaces for targeted marketing associated with a population of assets | |
CN106779074A (en) | Market brand combination prediction method and prediction server | |
CN108572988A (en) | A kind of house property assessment data creation method and device | |
Nilsson | The influence of urban and natural amenities on second home prices | |
Borgoni et al. | The value of culture to urban housing markets | |
CN109885745A (en) | A kind of user draws a portrait method, apparatus, readable storage medium storing program for executing and terminal device | |
d’Amato | A brief outline of AVM models and standards evolutions | |
E. Simlai | Predicting owner-occupied housing values using machine learning: an empirical investigation of California census tracts data | |
Tang et al. | Forecasting Airbnb prices through machine learning | |
Nilsson | Are valuations of place-based amenities driven by scale? | |
Bogdanova et al. | A wavelet-based approach to the analysis and modelling of financial time series exhibiting strong long-range dependence: the case of Southeast Europe | |
Efthymiou et al. | Modeling the impact of large-scale transportation infrastructure development on land cover | |
Rajballie et al. | Water consumption forecasting models–a case study in Trinidad (Trinidad and Tobago) | |
Kim et al. | How the pattern recognition ability of deep learning enhances housing price estimation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |