CN109544215A - Value of house prediction technique, device, computer equipment and storage medium - Google Patents
Value of house prediction technique, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN109544215A CN109544215A CN201811289846.6A CN201811289846A CN109544215A CN 109544215 A CN109544215 A CN 109544215A CN 201811289846 A CN201811289846 A CN 201811289846A CN 109544215 A CN109544215 A CN 109544215A
- Authority
- CN
- China
- Prior art keywords
- index
- value
- house
- prediction
- extraction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
Landscapes
- Business, Economics & Management (AREA)
- Development Economics (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
This application involves smart city technical fields, applied to real estate industry, in particular to a kind of Value of house prediction technique, device, computer equipment and storage medium, wherein, method includes: according to historical data, single argument calibrating is carried out to data, accurately determine the best lag period of each index, rationally generate sample data set, and room rate prediction prediction model is accurately constructed based on default machine learning method, building room rate prediction prediction model can treat estimation range Value of house Accurate Prediction, its experience for being not necessarily to rely on appraiser and personal subjective judgement, based on the rigorous treatment process of existing historical data, it can be realized the Accurate Prediction to Value of house.
Description
Technical field
This application involves prediction electric powder predictions, more particularly to a kind of Value of house prediction technique, device, computer
Equipment and storage medium.
Background technique
In real life, room rate has become the focal point of people's daily life, and the variation of room rate affects each row
The heart of each industry and ordinary people is whether engaged in the professional of the industries such as development of real estate, Real Estate Finance and building
Or ordinary people is intended to can have a more accurately prediction prediction to the following room rate tendency.
Traditional room rate prediction majority is that the appraiser of profession is supplied based on the proximal segment time come some regional Basic Housing Price, the source of houses
Relationship, policy and experience is needed to provide room rate prediction.This mode can generally depend critically upon the subjective judgement of appraiser
And experience, for the room rate of the same area, different appraisers finally show that room rate prediction may be different.
It can be seen that traditional room rate prediction technique is in significant ASIC limitation, room rate prediction result is not accurate enough.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of Value of house prediction for capableing of Accurate Prediction room rate
Method, apparatus, computer equipment and storage medium.
A kind of Value of house prediction technique, which comprises
Value of house historical data in region to be predicted is obtained, extracting from the Value of house historical data influences house valence
The index and Value of house index of value；
Index and Value of house index to extraction carry out single argument calibrating, determine the best lag period of index；
The index that preset quantity is filtered out from the index of the extraction, according to the corresponding best lag of the index filtered out
Phase generates sample data set；
According to the sample data set, room rate prediction prediction model is constructed.
The index and the progress single argument calibrating of Value of house index of described pair of extraction in one of the embodiments, really
Before the best lag period for determining index, further includes:
The index of the extraction is standardized；
The index and the progress single argument calibrating of Value of house index of described pair of extraction, determine the packet of best lag period of index
It includes:
To the index and the progress single argument calibrating of Value of house index after standardization, the best lag of index is determined
Phase.
The index of described pair of extraction, which is standardized, in one of the embodiments, includes:
Rule is filled up according to preset missing values, missing values are carried out to the index that there is missing in the index and are filled up, are obtained
The data set finished is filled up to missing values；
The data set finished is filled up for missing values, and according to preset index frequency conversion rule, frequency-conversion processing is carried out to index；
According to the index after frequency-conversion processing, the corresponding derivative index of index is determined；
Index conversion, the derivative index and corresponding frequency-conversion processing after index is converted are carried out to the derivative index
Index afterwards merges, the index after obtaining standardization.
The preset missing values fill up rule in one of the embodiments, are as follows: miss rate are less than or equal to pre-
If the index of threshold value, according to index property and index deletion condition, filled up to missing values are carried out there are the index of missing values；For
The index that miss rate is greater than the preset threshold is rejected.
The single argument calibrating includes the calibrating of economic meanings, T calibrating and correlation inspection in one of the embodiments,
It is fixed；The index and the progress single argument calibrating of Value of house index of described pair of extraction, determine that the best lag period of index includes:
It is derivative to the index progress lag period of extraction, the index under the different lag periods is generated, and determine under the different lag periods
The corresponding Value of house index of index；
The correlation between the corresponding Value of house index of index under the different lag period is calculated, according to economy
Meaning and correlation carry out economic meanings calibrating to the index under the different lag periods；
The housing price index corresponding to the index under the different lag periods carries out T calibrating；
According to T verification result, each index related conspicuousness is judged, the best of each index is determined according to the conspicuousness
Lag period.
Described according to the sample data set in one of the embodiments, constructing room rate prediction prediction model includes:
Choosing the sample data and concentrating first part's data is training data, passes through multiple default machine learning sides respectively
The training of method prediction model constructs different room rate prediction prediction models；
It is described according to the sample data set, after constructing room rate prediction prediction model, further includes:
Choosing the sample data and concentrating second part data is test data, predicts mould to each default machine learning method
The room rate prediction prediction model that type training obtains is tested, and the smallest machine learning method prediction model pair of mean error is selected
The room rate prediction prediction model answered is optimal room rate prediction prediction model.
The index that preset quantity is filtered out from the index of the extraction in one of the embodiments, according to sieve
The index the selected corresponding best lag period, generating sample data set includes:
According to preset shortlist create-rule, the index generation that preset quantity is filtered out from the index of the extraction is short
List；
According to the corresponding best lag period of index in shortlist and generate sample data set.
A kind of Value of house prediction meanss, described device include:
Extraction module, for obtaining Value of house historical data in region to be predicted, from the Value of house historical data
Extract the index and Value of house index for influencing Value of house；
Single argument assay module, it is determining to refer to for the index and the progress single argument calibrating of Value of house index to extraction
The target best lag period；
Sample data set generation module, for filtering out the index of preset quantity from the index of the extraction, according to sieve
The index the selected corresponding best lag period generates sample data set；
Model construction module, for constructing room rate prediction prediction model according to the sample data set.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device is realized when executing the computer program such as the step of the above method.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
It realizes when row such as the step of above-mentioned method.
Premises Value Prediction Methods, device, computer equipment and storage medium, according to historical data, to data into
The calibrating of row single argument, accurately determines the best lag period of each index, rationally generates sample data set, and based on default machine learning
Method accurately constructs room rate prediction prediction model, and it is accurate that building room rate prediction prediction model can treat estimation range Value of house
Prediction, without relying on the experience and personal subjective judgement of appraiser, based on the existing rigorous treatment process of historical data, Neng Goushi
Now to the Accurate Prediction of Value of house.
Detailed description of the invention
Fig. 1 is the applied environment figure of Value of house prediction technique in one embodiment；
Fig. 2 is the flow diagram of Value of house prediction technique in one embodiment；
Fig. 3 is the flow diagram of Value of house prediction technique in another embodiment；
Fig. 4 is the structural block diagram of Value of house prediction meanss in one embodiment；
Fig. 5 is the structural block diagram of Value of house prediction meanss in another embodiment；
Fig. 6 is the experimental result comparison diagram using premises Value Prediction Methods；
Fig. 7 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
Value of house prediction technique provided by the present application, can be applied in application environment as shown in Figure 1.Wherein, eventually
End 102 is communicated with server 104 by network by network.User passes through terminal 102 for each region of daily collection
Value of house data are sent to server 104, which includes the index and Value of house index of Value of house,
Server 104 constructs database, which is used to specially store the corresponding region Value of house number that each terminal 102 uploads
According to.When needing to carry out Value of house prediction prediction, terminal 102 sends regional location data to be predicted to server 104, service
Device 104 obtains Value of house historical data in region to be predicted, and the finger for influencing Value of house is extracted from Value of house historical data
It is marked with and Value of house index；Index and Value of house index to extraction carry out single argument calibrating, determine the best of index
Lag period；The index that preset quantity is filtered out from the index of extraction, it is raw according to the index filtered out the corresponding best lag period
At sample data set；According to sample data set, room rate prediction prediction model is constructed；Prediction model is looked forward to the prospect to be predicted according to room rate
Region Value of house is predicted, Value of house prediction result is obtained, and sends Value of house prediction result to terminal 102.Wherein,
Terminal 102, which can be, but not limited to, to be various personal computers, laptop, smart phone, tablet computer and portable wears
Equipment is worn, server 104 can be realized with the server cluster of the either multiple server compositions of independent server.
In one embodiment, as shown in Fig. 2, providing a kind of Value of house prediction technique, it is applied to Fig. 1 in this way
In server for be illustrated, comprising the following steps:
S200: obtaining Value of house historical data in region to be predicted, and extracting from Value of house historical data influences house
The index and Value of house index of value.
Region to be predicted refers to that the target area of this Value of house prediction, the region can be some administrative region,
Such as Beijing, Shanghai, Guangzhou etc..The region can also be a smaller range, such as some cell etc..Region house to be predicted
Value historical data can be the terminal in current entry and acquire the data being sent under server aggregates, can be server
Outside, which is obtained, by means such as internets has corresponding data.Extracting in Value of house historical data influences Value of house
Index and Value of house index, the index for influencing Value of house includes: all kinds of macro-performance indicators, for example, GDP, CPI, PMI,
Per capita disposable income etc.；Meso-economics index, for example, each city (area) Urbanization Rate, subway mileage, per capita living space with
And the commercial house area for sale etc.；Policies and regulations such as real estate limit is sold limit purchase policy, first suite interest rate policy, is sent out in city for a long time
Exhibition planning etc..Value of house index specifically can be room rate, may include hanging disk and transaction value.It is non-essential, in order to true
Protect the accuracy of subsequent Value of house prediction, the Value of house historical data in region to be predicted in the available proximal segment time, example
It such as obtains nearest 1 year, obtain the Value of house historical data in region to be predicted in the nearest 6 months time, for the number of acquisition
It rationally arranges corresponding index according to the time is also based on, such as using the moon as foundation.Such as by taking " resident population " this index as an example,
The history value of the index be [h1, h2 ..., hi ...], wherein hi indicates i-th month resident population within a preset time
Number.
S400: index and Value of house index to extraction carry out single argument calibrating, determine the best lag period of index.
The lag period refers to that just some influences the data of Value of house and can reflect after lagging certain time, in order to standard
Really prediction Value of house, it is thus necessary to determine that the best lag period of good index.Single argument examines and determine the inspection that can specifically include economic meanings
Fixed, T calibrating and correlation calibrating.
S600: filtering out the index of preset quantity from the index of extraction, corresponding best stagnant according to the index filtered out
Later period generates sample data set.
A certain number of indexs are filtered out from the index of extraction, and corresponding best according to the index that these are filtered out
Lag period generates sample data set.It is non-essential, it can be based on preset shortlist create-rule, sieved from the index of extraction
A certain number of indexs are selected, according to the index filtered out the corresponding best lag period, generate sample data set.Preset short name
Single create-rule is the inventory index for having the model training that model determines based on real estate industry's expertise, real estate.
S800: according to sample data set, room rate prediction prediction model is constructed.
Specifically, room rate prediction prediction model can be constructed by machine learning method.Default machine learning method can
To include linear regression, Lasso, ridge regression (Ridge Regression), random forest, k nearest neighbor algorithm (k Neighbour
Regression), decision tree, Support vector regression (SVR), grad enhancement return
(GradientBoostingRegressor) model and XGBoost algorithm, using the sample data set that step S600 is obtained as
Training data can construct corresponding room rate prediction prediction mould by any one of the above machine learning method prediction model
Type.
Premises Value Prediction Methods carry out single argument calibrating to data, accurately determine each index according to historical data
The best lag period, rationally generate sample data set, and room rate prediction prediction mould is accurately constructed based on default machine learning method
Type, building room rate prediction prediction model can treat estimation range Value of house Accurate Prediction, without relying on the warp of appraiser
It tests and the Accurate Prediction to Value of house can be realized based on the existing rigorous treatment process of historical data with personal subjective judgement.
As shown in figure 3, in one of the embodiments, before step S400, further includes:
S300: the index of extraction is standardized；Step S400 specifically: to the index after standardization with
And Value of house index carries out single argument calibrating, determines the best lag period of index.
The index of extraction is standardized, standardization mainly includes removal exceptional value, trend and seasonality
Influence processing.Index after being standardized more can objective characterisation Value of house variation tendency, be conducive to subsequent accurate structure
Building valence prediction prediction model.
More specifically, the index of extraction is standardized and includes:
Step 1: filling up rule according to preset missing values, carries out missing values to the index that there is missing in index and fills up,
It obtains missing values and fills up the data set finished.
Certain indexs the case where there are shortage of data, in this case according to preset missing values fill up rule with
And data with existing carries out missing values tune benefit, polishing data set.Specifically, the finger of preset threshold is less than or equal to for miss rate
Mark is filled up according to index property and index deletion condition to missing values are carried out there are the index of missing values；Miss rate is greater than
The index of preset threshold is rejected.In practical applications, for pre-set level of the miss rate less than or equal to 30%, root
According to index property and index deletion condition, filled up to missing values are carried out there are the index of missing values；And miss rate is greater than
For 30% pre-set level, (in the case where investigating remaining available data source can not fill up) picks the index
It removes.When factor depletion be index periodically lack, such as annual January, 2 month data periodically lack.Due to the missing feelings
Condition is related to statistics bureau's statistical work period, therefore, is not fixed the influence of factor bring to eliminate the date in the Spring Festival, enhances data
Comparativity, certain index in January, 2 months need to be filled up.If the index is aggregate-value, with number in March current year
According to one third, 2/3rds make respectively this year January, 2 month shortage of data value fill up；If the index is this month
Value, then with 3 month value of year make current year 1,2 month missing values fill up.When the index missing number of factor depletion is less, irregular
Property, if the index is aggregate-value, linear interpolation is carried out according to the latter moon data before missing this month and fills up missing；If the index
Value was actually occurred for this month, then is filled up with distance missing nearest 6 months of the moon.For the special index in part, such as construction surface
Product, due to the particularity of the index property, retrodicts missing values using the mean annual rate of increase.
Step 2: filling up the data set finished for missing values, according to preset index frequency conversion rule, becomes to index
Frequency is handled.
Monthly data is converted by the method for linear interpolation by the index in season, year, realizes the frequency conversion of pre-set level
Processing, convenient for the derivative index of subsequent calculating.For example, " GDP " this index is season data, " permanent resident population " this index is year
Degree evidence carries out linear interpolation usually using the annual historical data of continuous two season or two, every month is calculated
Data.
Step 3: according to the index after frequency-conversion processing, the corresponding derivative index of index is determined.
The general relevant derivative index being related to by subsystem of Value of house is 24 total, can directly obtain from data source
Total 13, remaining 11 each derivative indexs, which mainly the methods of are divided by, are subtracted each other by certain existing several index, to be obtained.Such as: it is " permanent
This index of population/household registration population's ratio " is obtained by " permanent resident population " and " household registration population " the two indexs derivative.
Step 4: index conversion, derivative index and corresponding frequency-conversion processing after index is converted are carried out to derivative index
Index afterwards merges, the index after obtaining standardization.
Derivative index generation finishes, that is, forms the wide table of data set before index converts.Based on this, then makees index to it and turn
Change, index transform mode includes: that 3 on a month-on-month basis, a year-on-year, standardization and original value.For example, referring to for room trading volume
Mark, will use 3 it is on a month-on-month basis, accumulation Value Data (for example, sale area) will use a year-on-year, index sheet is as ratio, meeting
Use original value.It should be noted that in index conversion process, the index that need to partially convert on year-on-year basis, since initial data rises
The limitation of time beginning, it may appear that after conversion the case where shortage of data, after such index missing can be converted with index in data
Digit is filled up as missing values.
Single argument calibrating includes the calibrating of economic meanings, T calibrating and correlation calibrating in one of the embodiments,.
Specifically, the calibrating of economic meanings specifically: the stock index of index will be combined to be judged.It is specific
If including: that the economic meanings of certain index are positive, i.e., the index value is bigger, has facilitation to second-hand house price, conversely, if
Economic meanings are negative, then the index value is bigger, inhibited to second-hand house price, for example, trading volume and room rate are just
Correlation, the positive correlation of M2 (while reflecting reality and potential purchasing power) and room rate, interest rate and room rate are negatively correlated.By comparing
The direction of the index related coefficient and the same tropism of economic meanings, to judge whether the index passes through the calibrating of economic meanings.T inspection
Fixed judgment criteria are as follows: if T calibrating p value be less than or equal to 0.05, the index by T examine and determine, it is on the contrary then determine T examine and determine it is obstructed
It crosses.T inspection is one kind of univariate analysis, and index such as GDP and growth rate of real estate price are carried out T inspection, see if there is correlation, if
Pass through inspection, it is determined whether use this index.The purpose of the step is, closes according to the T p value examined and index meaning are selected
Suitable index.For a step, according to the inspection result of different lag periods, the correlation most significant lag period is selected, if phase
Closing property is not significant, does not just select this index.The lag period of index is determined according to statistic p value, p≤
0.05 is significant.By calculating any index 3 months, 6 months, 9 months, 12 months predictive abilities to room rate were determined most
The excellent prediction lag period.It is selected by single argument calibrating and best lag period, that is, it can determine each index with the presence or absence of best
Lag period.If index there are the best lag period, takes the corresponding best lag period data training pattern of the index, if index is related to
All lag periods it is not significant, then specifying the best lag period of such index is March (3 months models of prediction) or December
(12 months models of prediction).
As shown in figure 3, step S600 includes: in one of the embodiments,
S620: it is derivative to the index progress lag period of extraction, the index under the different lag periods is generated, and determine different lag
The corresponding Value of house index of index under phase.
S640: the correlation between the corresponding Value of house index of index under the different lag periods is calculated, according to warp
Meaning of helping and correlation carry out economic meanings calibrating to the index under the different lag periods.
Here correlation can be using the phase of housing price index income time series and the index time series adjusted through lag
Relationship number characterization.Specific economic meanings calibrating are as follows: when economic meanings are in the same direction with correlation, judge that economic meanings calibrating passes through,
When economic meanings are reversed with correlation, economic meanings calibrating failure is judged.
S660: the housing price index corresponding to the index under the different lag periods carries out T calibrating.
The housing price index corresponding to the index under the different lag periods carries out T calibrating, when p is less than or equal to 0.05, sentences
Disconnected T calibrating passes through, and when p is greater than 0.05, judges T calibrating failure.
S680: according to T verification result, judge each index related conspicuousness, each index is determined most according to conspicuousness
The good lag period.
According to the size of p value, each index related conspicuousness is judged, the best lag of each index is determined according to conspicuousness
Phase.Based on wide table is modeled, each explanatory variable (above-mentioned all indexs) progress lag period is derived, different lag period (n=are generated
3,6,12,15,18,24) index under substitutes into univariate analysis respectively and carries out single argument calibrating, most has explanation strengths to find
Index, and determine its best lag period (p value is less than or equal to 0.05).
As shown in figure 3, step S800 includes: to choose sample data to concentrate first part's number in one of the embodiments,
Different room rate prediction predictions is constructed respectively by multiple default machine learning method prediction model training according to for training data
Model；Further include step S900 after step S800: choosing sample data and concentrating second part data is test data, to each
The room rate prediction prediction model that default machine learning method prediction model training obtains is tested, and selects mean error the smallest
The corresponding room rate prediction prediction model of machine learning method prediction model is optimal room rate prediction prediction model.
Machine learning method includes linear regression, Lasso, ridge regression (Ridge Regression), random forest, K close
Adjacent algorithm (k Neighbour Regression), decision tree, Support vector regression (SVR), grad enhancement return
(GradientBoostingRegressor) model and XGBoost algorithm, different room rates can be constructed based on these algorithms
Prediction prediction model, first part's data test each room rate prediction prediction model as test data using in sample data
Mean error, selecting the corresponding room rate prediction prediction model of the smallest machine learning method prediction model of mean error is optimal room
Valence prediction prediction model.
The index for filtering out preset quantity from the index of extraction in one of the embodiments, according to the finger filtered out
It marks the corresponding best lag period, generating sample data set includes: according to preset shortlist create-rule, from the index of extraction
The index for filtering out preset quantity generates shortlist；According to the corresponding best lag period of index in shortlist and generate sample data
Collection.
Shortlist create-rule can be obtained based on historical empirical data, specifically real estate industry expert be combined to pass through
Test and have real estate models discussion generation.According to the shortlist create-rule, present count is filtered out from the index of extraction
The index of amount generates shortlist, such as can choose significance level in 53 indexs is high index as model training short name
It is single, sample data set is generated according to the best lag period data of each index, for machine learning modeling training.It may be noted that
It is that sample data concentration includes training data and test data, and training data is for machine learning modeling training, test data
For testing whether established model is predicted accurately.
More specifically, choosing sample data and concentrating second part data is test data, to each default machine learning
The room rate prediction prediction model that the training of method prediction model obtains is tested, and the smallest machine learning method of mean error is selected
The corresponding room rate prediction prediction model of prediction model is that optimal room rate prediction prediction model specifically includes:
D1, configuration is grouped to all indexs in shortlist, according to grouping situation, is successively obtained from sample data concentration
Take it is each grouping it is corresponding enter modular character training set, test set.
D2, using it is each grouping it is corresponding enter modular character training set, preset machine learning method is trained, structure
Building valence look-forward model.
D3, using it is each grouping it is corresponding enter modular character test set, to the corresponding room rate of each machine learning method look forward to the prospect
The accuracy of model is tested.
D4, the mean error (RMSE) for calculating the corresponding test result of each room rate look-forward model choose mean error
(RMSE) the corresponding room rate prediction prediction model of the smallest algorithm is as optimal room rate look-forward model.
Firstly, be grouped configuration to shortlist, each grouping enter the control of modular character quantity one and only one, by
In different cities, its quality of data is not quite similar, if all equal no datas of index in being grouped, this group of index quantity is zero.Example
Such as, it during packet configuration, is grouped according to pointer type: middle sight, macroscopic view, derivative etc..Wherein, training pattern combines
Quantity is that the traversal of 1 index is chosen in all groupings.For example, B group has 2 indexs, then number of combinations is if A group has 3 indexs
3*2=6, totally 6 kinds, combined index has 2.Based on all number of combinations of model, 9 kinds of engineerings will be respectively adopted to each combination
Learning method is trained, and is respectively as follows: linear regression, Lasso, ridge regression (Ridge Regression), random forest, k nearest neighbor
Algorithm (k Neighbour Regression), decision tree, Support vector regression (SVR), grad enhancement return
(GradientBoostingRegressor) model and XGBoost algorithm.By the training of the above method, average miss is chosen
The corresponding room rate prediction prediction model of difference (RMSE) the smallest algorithm is as optimal room rate look-forward model.
It should be understood that although each step in the flow chart of Fig. 2-3 is successively shown according to the instruction of arrow,
These steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly stating otherwise herein, these steps
Execution there is no stringent sequences to limit, these steps can execute in other order.Moreover, at least one in Fig. 2-3
Part steps may include that perhaps these sub-steps of multiple stages or stage are not necessarily in synchronization to multiple sub-steps
Completion is executed, but can be executed at different times, the execution sequence in these sub-steps or stage is also not necessarily successively
It carries out, but can be at least part of the sub-step or stage of other steps or other steps in turn or alternately
It executes.
As shown in figure 4, a kind of Value of house prediction meanss, device include:
Extraction module 200 is mentioned from Value of house historical data for obtaining Value of house historical data in region to be predicted
Take the index and Value of house index for influencing Value of house；
Single argument assay module 400 is determined for the index and the progress single argument calibrating of Value of house index to extraction
The best lag period of index；
Sample data set generation module 600, for filtering out the index of preset quantity from the index of extraction, according to screening
The corresponding best lag period of index out generates sample data set；
Model construction module 800, for constructing room rate prediction prediction model according to sample data set.
Premises value forecasting device carries out single argument calibrating to data, accurately determines each index according to historical data
The best lag period, rationally generate sample data set, and room rate prediction prediction mould is accurately constructed based on default machine learning method
Type, building room rate prediction prediction model can treat estimation range Value of house Accurate Prediction, without relying on the warp of appraiser
It tests and the Accurate Prediction to Value of house can be realized based on the existing rigorous treatment process of historical data with personal subjective judgement.
Premises value forecasting device in one of the embodiments, further include: processing module, for the finger to extraction
Mark is standardized；Single argument assay module 400 is also used to the index and Value of house index after standardization
Single argument calibrating is carried out, determines the best lag period of index.
Processing module is also used to fill up rule according to preset missing values in one of the embodiments, to depositing in index
Missing values are carried out in the index of missing to fill up, and are obtained missing values and are filled up the data set finished；The number finished is filled up for missing values
Frequency-conversion processing is carried out to index according to preset index frequency conversion rule according to collection；According to the index after frequency-conversion processing, index is determined
Corresponding derivative index；Index conversion, derivative index and corresponding frequency-conversion processing after index is converted are carried out to derivative index
Index afterwards merges, the index after obtaining standardization.
Preset missing values fill up rule in one of the embodiments, are as follows: are less than or equal to default threshold for miss rate
The index of value is filled up according to index property and index deletion condition to missing values are carried out there are the index of missing values；For missing
The index that rate is greater than preset threshold is rejected.
Single argument calibrating includes the calibrating of economic meanings, T calibrating and correlation calibrating in one of the embodiments,；It is single
Variable assay module 400 is also used to derive the index progress lag period of extraction, generates the index under the different lag periods, and determine
The corresponding Value of house index of index under the different lag periods；Calculate the corresponding Value of house of index under the different lag periods
Correlation between index carries out economic meanings calibrating to the index under the different lag periods according to economic meanings and correlation；It is right
The corresponding housing price index of index under the different lag periods carries out T calibrating；According to T verification result, judge each index related
Conspicuousness, the best lag period of each index is determined according to conspicuousness.
As shown in figure 5, model construction module 800 is also used to choose sample data and concentrates the in one of the embodiments,
A part of data are training data, respectively by multiple default machine learning method prediction model training, construct different room rates
Prediction prediction model.Premises value forecasting device further includes preferred module 900, concentrates second for choosing sample data
Divided data is test data, and the room rate prediction prediction model obtained to each default machine learning method prediction model training is surveyed
Examination selects the corresponding room rate prediction prediction model of the smallest machine learning method prediction model of mean error to look forward to the prospect for optimal room rate
Prediction model.
Sample data set generation module 600 is also used to be generated according to preset shortlist and advise in one of the embodiments,
Then, the index that preset quantity is filtered out from the index of extraction generates shortlist；It is corresponding best stagnant according to index in shortlist
Later period simultaneously generates sample data set.
Specific about Value of house prediction meanss limits the limit that may refer to above for Value of house prediction technique
Fixed, details are not described herein.Modules in premises value forecasting device can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
In practical application, a certain region room rate in Chongqing is predicted with the room rate prediction prediction model of the application building,
Shown in its obtained experimental result Fig. 6.It can be accurately to a certain area in Chongqing based on the visible the application room rate prediction prediction model of Fig. 6
Domain room rate is predicted.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition can be as shown in Figure 7.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing machine learning method data.The network interface of the computer equipment is used for and external terminal
It is communicated by network connection.To realize a kind of Value of house prediction technique when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Fig. 7, only part relevant to application scheme is tied
The block diagram of structure does not constitute the restriction for the computer equipment being applied thereon to application scheme, specific computer equipment
It may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment is provided, including memory, processor and storage are on a memory
And the computer program that can be run on a processor, processor perform the steps of when executing computer program
Value of house historical data in region to be predicted is obtained, extracting from Value of house historical data influences Value of house
Index and Value of house index；
Index and Value of house index to extraction carry out single argument calibrating, determine the best lag period of index；
The index that preset quantity is filtered out from the index of extraction, according to the index filtered out the corresponding best lag period,
Generate sample data set；
According to sample data set, room rate prediction prediction model is constructed.
In one embodiment, it is also performed the steps of when processor executes computer program
The index of extraction is standardized.
In one embodiment, it is also performed the steps of when processor executes computer program
Rule is filled up according to preset missing values, missing values are carried out to the index that there is missing in index and are filled up, are lacked
Mistake value fills up the data set finished；The data set finished is filled up for missing values, according to preset index frequency conversion rule, to index
Carry out frequency-conversion processing；According to the index after frequency-conversion processing, the corresponding derivative index of index is determined；Index is carried out to derivative index to turn
Change, the index after derivative index and corresponding frequency-conversion processing after index is converted merges, after obtaining standardization
Index.
In one embodiment, it is also performed the steps of when processor executes computer program
It is derivative to the index progress lag period of extraction, the index under the different lag periods is generated, and determine under the different lag periods
The corresponding Value of house index of index；Calculate the phase between the corresponding Value of house index of index under the different lag periods
Guan Xing carries out economic meanings calibrating to the index under the different lag periods according to economic meanings and correlation；To under the different lag periods
The corresponding housing price index of index carry out T calibrating；According to T verification result, each index related conspicuousness is judged, according to
Conspicuousness determines the best lag period of each index.
In one embodiment, it is also performed the steps of when processor executes computer program
Choosing sample data and concentrating first part's data is training data, pre- by multiple default machine learning methods respectively
Model training is surveyed, different room rate prediction prediction models is constructed；Choosing sample data and concentrating second part data is test data,
The room rate prediction prediction model obtained to each default machine learning method prediction model training is tested, and selects mean error most
The corresponding room rate prediction prediction model of small machine learning method prediction model is optimal room rate prediction prediction model.
In one embodiment, it is also performed the steps of when processor executes computer program
According to preset shortlist create-rule, the index that preset quantity is filtered out from the index of extraction generates short name
It is single；According to the corresponding best lag period of index in shortlist and generate sample data set.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Machine program performs the steps of when being executed by processor
Value of house historical data in region to be predicted is obtained, extracting from Value of house historical data influences Value of house
Index and Value of house index；
Index and Value of house index to extraction carry out single argument calibrating, determine the best lag period of index；
The index that preset quantity is filtered out from the index of extraction, according to the index filtered out the corresponding best lag period,
Generate sample data set；
According to sample data set, room rate prediction prediction model is constructed.
In one embodiment, it is also performed the steps of when computer program is executed by processor
The index of extraction is standardized.
In one embodiment, it is also performed the steps of when computer program is executed by processor
Rule is filled up according to preset missing values, missing values are carried out to the index that there is missing in index and are filled up, are lacked
Mistake value fills up the data set finished；The data set finished is filled up for missing values, according to preset index frequency conversion rule, to index
Carry out frequency-conversion processing；According to the index after frequency-conversion processing, the corresponding derivative index of index is determined；Index is carried out to derivative index to turn
Change, the index after derivative index and corresponding frequency-conversion processing after index is converted merges, after obtaining standardization
Index.
In one embodiment, it is also performed the steps of when computer program is executed by processor
It is derivative to the index progress lag period of extraction, the index under the different lag periods is generated, and determine under the different lag periods
The corresponding Value of house index of index；Calculate the phase between the corresponding Value of house index of index under the different lag periods
Guan Xing carries out economic meanings calibrating to the index under the different lag periods according to economic meanings and correlation；To under the different lag periods
The corresponding housing price index of index carry out T calibrating；According to T verification result, each index related conspicuousness is judged, according to
Conspicuousness determines the best lag period of each index.
In one embodiment, it is also performed the steps of when computer program is executed by processor
Choosing sample data and concentrating first part's data is training data, pre- by multiple default machine learning methods respectively
Model training is surveyed, different room rate prediction prediction models is constructed；Choosing sample data and concentrating second part data is test data,
The room rate prediction prediction model obtained to each default machine learning method prediction model training is tested, and selects mean error most
The corresponding room rate prediction prediction model of small machine learning method prediction model is optimal room rate prediction prediction model.
In one embodiment, it is also performed the steps of when computer program is executed by processor
According to preset shortlist create-rule, the index that preset quantity is filtered out from the index of extraction generates short name
It is single；According to the corresponding best lag period of index in shortlist and generate sample data set.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Instruct relevant hardware to complete by computer program, computer program to can be stored in a non-volatile computer readable
It takes in storage medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein, this Shen
Please provided by any reference used in each embodiment to memory, storage, database or other media, may each comprise
Non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
Above embodiments only express the several embodiments of the application, and the description thereof is more specific and detailed, but can not
Therefore it is construed as limiting the scope of the patent.It should be pointed out that for those of ordinary skill in the art,
Under the premise of not departing from the application design, various modifications and improvements can be made, these belong to the protection scope of the application.
Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (10)
1. a kind of Value of house prediction technique, which comprises
Value of house historical data in region to be predicted is obtained, extracting from the Value of house historical data influences Value of house
Index and Value of house index；
Index and Value of house index to extraction carry out single argument calibrating, determine the best lag period of index；
The index that preset quantity is filtered out from the index of the extraction, according to the index filtered out the corresponding best lag period,
Generate sample data set；
According to the sample data set, room rate prediction prediction model is constructed.
2. the method according to claim 1, wherein the index and the progress of Value of house index of described pair of extraction
Single argument calibrating, before the best lag period for determining index, further includes:
The index of the extraction is standardized；
The index and the progress single argument calibrating of Value of house index of described pair of extraction, determine that the best lag period of index includes:
To the index and the progress single argument calibrating of Value of house index after standardization, the best lag period of index is determined.
3. according to the method described in claim 2, it is characterized in that, the index of described pair of extraction is standardized and includes:
Rule is filled up according to preset missing values, missing values are carried out to the index that there is missing in the index and are filled up, are lacked
Mistake value fills up the data set finished；
The data set finished is filled up for missing values, and according to preset index frequency conversion rule, frequency-conversion processing is carried out to index；
According to the index after frequency-conversion processing, the corresponding derivative index of index is determined；
To the derivative index progress index conversion, after the derivative index and corresponding frequency-conversion processing after index is converted
Index merges, the index after obtaining standardization.
4. according to the method described in claim 3, it is characterized in that, the preset missing values fill up rule are as follows: for missing
Rate is less than or equal to the index of preset threshold, according to index property and index deletion condition, to there are the progress of the index of missing values
Missing values are filled up；The index for being greater than the preset threshold for miss rate is rejected.
5. the method according to claim 1, wherein single argument calibrating includes the calibrating of economic meanings, T inspection
The calibrating of fixed and correlation；The index and the progress single argument calibrating of Value of house index of described pair of extraction, determine the best stagnant of index
Later period includes:
It is derivative to the index progress lag period of extraction, the index under the different lag periods is generated, and determine the finger under the different lag periods
Mark corresponding Value of house index；
The correlation between the corresponding Value of house index of index under the different lag period is calculated, according to economic meanings
And correlation, economic meanings calibrating is carried out to the index under the different lag periods；
The housing price index corresponding to the index under the different lag periods carries out T calibrating；
According to T verification result, each index related conspicuousness is judged, the best lag of each index is determined according to the conspicuousness
Phase.
6. the method according to claim 1, wherein described according to the sample data set, building room rate prediction
Prediction model includes:
Choosing the sample data and concentrating first part's data is training data, pre- by multiple default machine learning methods respectively
Model training is surveyed, different room rate prediction prediction models is constructed；
It is training data that the selection sample data, which concentrates first part's data, passes through multiple default machine learning sides respectively
Method prediction model training, after constructing different room rate prediction prediction models, further includes:
Choosing the sample data and concentrating second part data is test data, is instructed to each default machine learning method prediction model
The room rate prediction prediction model got is tested, and selects the smallest machine learning method prediction model of mean error corresponding
Room rate looks forward to the prospect prediction model as optimal room rate prediction prediction model.
7. the method according to claim 1, wherein described filter out preset quantity from the index of the extraction
Index, according to the index filtered out the corresponding best lag period, generating sample data set includes:
According to preset shortlist create-rule, the index that preset quantity is filtered out from the index of the extraction generates short name
It is single；
According to the corresponding best lag period of index in shortlist and generate sample data set.
8. a kind of Value of house prediction meanss, which is characterized in that described device includes:
Extraction module is extracted from the Value of house historical data for obtaining Value of house historical data in region to be predicted
Influence the index and Value of house index of Value of house；
Single argument assay module determines index for the index and the progress single argument calibrating of Value of house index to extraction
The best lag period；
Sample data set generation module, for filtering out the index of preset quantity from the index of the extraction, according to filtering out
The index corresponding best lag period, generate sample data set；
Model construction module, for constructing room rate prediction prediction model according to the sample data set.
9. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 7 the method when executing the computer program.
10. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any one of claims 1 to 7 is realized when being executed by processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811289846.6A CN109544215A (en) | 2018-10-31 | 2018-10-31 | Value of house prediction technique, device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811289846.6A CN109544215A (en) | 2018-10-31 | 2018-10-31 | Value of house prediction technique, device, computer equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109544215A true CN109544215A (en) | 2019-03-29 |
Family
ID=65846274
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811289846.6A Pending CN109544215A (en) | 2018-10-31 | 2018-10-31 | Value of house prediction technique, device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109544215A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
-
2018
- 2018-10-31 CN CN201811289846.6A patent/CN109544215A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109376936A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| US20030195830A1 (en) | System, method and framework for generating scenarios | |
| Minku et al. | How to make best use of cross-company data for web effort estimation? | |
| Lazzari et al. | User behaviour models to forecast electricity consumption of residential customers based on smart metering data | |
| CN104978479B (en) | For predicting the device of water requirement | |
| JP7304698B2 (en) | Water demand forecasting method and system | |
| Zimmermann et al. | Pathways to truth: comparing different upscaling options for an agent-based sector model | |
| CN114398669A (en) | Joint credit scoring method and device based on privacy protection calculation and cross-organization | |
| CN109635996A (en) | Room rate tendency look-forward model construction method, device and storage medium | |
| CN109345035A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| CN109242578A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| CN114202179A (en) | Target enterprise identification method and device | |
| CN109544215A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| CN117853238A (en) | Power transaction auxiliary decision-making system based on multi-data source fusion | |
| Solmaz et al. | How external factors affect domestic economy: Nowcasting an emerging market | |
| CN109325811A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| Hoekstra et al. | Reducing the variation of environmental footprint estimates based on multiregional input–output databases | |
| CN109543904A (en) | Medical data prediction technique, device, computer equipment and storage medium | |
| CN112434019A (en) | Historical electric quantity tracing and cleaning method applied to household variable relation change and electric power center | |
| CN109492806A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| Lv et al. | Corn price prediction in China’s futures market during COVID-19 | |
| CN109214603A (en) | Value of house prediction technique, device, computer equipment and storage medium | |
| Los et al. | Neural network technology forecasting the country's business climate | |
| Amega et al. | A System Dynamics Modelling of a Long-term Residential Electricity Consumption in Lomé, Togo | |
| CN109492805A (en) | Value of house prediction technique, device, computer equipment and storage medium |
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: 20190329 |