CN107633421A - A kind of processing method and processing device of market prediction data - Google Patents
A kind of processing method and processing device of market prediction data Download PDFInfo
- Publication number
- CN107633421A CN107633421A CN201710735073.9A CN201710735073A CN107633421A CN 107633421 A CN107633421 A CN 107633421A CN 201710735073 A CN201710735073 A CN 201710735073A CN 107633421 A CN107633421 A CN 107633421A
- Authority
- CN
- China
- Prior art keywords
- training set
- prediction
- collection
- model
- obtains
- 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
Abstract
The invention discloses a kind of processing method and processing device of market prediction data, method therein includes:Using the tentative prediction data set being obtained ahead of time as the first training set;First training set is classified using convolutional neural networks model, and sorted first training set is predicted, obtains the first prediction result;First prediction result is added into sorted first training set, obtains the second training set;According to second training set, multiple target prediction collection are obtained respectively using bagging method, and market prediction is carried out based on the multiple target prediction collection.The present invention solves method of the prior art and the inaccurate technical problem of prediction result be present because forecast model is single.
Description
Technical field
The invention belongs to market prediction technical field, more particularly to a kind of processing method and processing device of market prediction data.
Background technology
With the improvement of people ' s living standards with growing changes in demand, the demand in market is also constantly changed, made
The direction of sale and species, which must be held, becomes more and more difficult, and how to carry out market prediction exactly becomes more and more important.
Current market prediction method, quantitative analysis is typically carried out according to sales volume, then draws forecast model, then
Market is predicted according to forecast model, but the forecast model drawn in method be only according to the sales volume of commodity come
Analyzed, do not consider other influence factors, thus existed and show that forecast model is more single, hence in so that prediction result
It is inaccurate.
It can be seen that there is the inaccurate technical problem of prediction result in method of the prior art because forecast model is single.
The content of the invention
The invention provides a kind of processing method and processing device of market prediction data, to solve method of the prior art
The inaccurate technical problem of prediction result be present because forecast model is single.
First aspect present invention provides a kind of processing method of market prediction data, and methods described includes:
Gather first behavioural information of the user to article;
According to first behavioural information, the feature related to purchase article is extracted;
According to the feature related to purchase article, prediction sets are built;
Based on the prediction sets, market is predicted.
Alternatively, described according to first behavioural information, before extracting the feature related to purchase article, the side
Method also includes:
First behavioural information is pre-processed, obtains the second behavioural information.
Alternatively, the behavioural information includes click behavior, buying behavior, collection behavior, addition shopping cart behavior,
It is described according to first behavioural information, the extraction feature related with buying article, including:
Respectively according to the click behavior, buying behavior, collection behavior and addition shopping cart behavior, counting user is default
Number of clicks, purchase number, collection number and addition shopping cart number in time;
According to the number of clicks, purchase number, collection number and shopping cart number is added, obtains the friendship of user and article
The mutual purchase ratio of purchase conversion ratio and user to article of rate, user to article.
Alternatively, in feature related to purchase article described in the basis, after building prediction sets, methods described is also
Including:
Based on preset rules, the prediction sets are updated.
Based on same inventive concept, second aspect of the present invention provides a kind of processing unit of market prediction data, institute
Stating device includes:
First obtains module, for the tentative prediction data set that will be obtained ahead of time as the first training set;
First prediction module, for being classified using convolutional neural networks model to first training set, and to dividing
The first training set after class is predicted, and obtains the first prediction result;
Second obtains module, for first prediction result to be added into sorted first training set, obtains
Obtain the second training set;
Second prediction module, for according to second training set, obtaining multiple target predictions respectively using bagging method
Collection, and market prediction is carried out based on the multiple target prediction collection.
Optionally, described device also includes processing module, for according to second training set, being divided using bagging method
After not obtaining multiple target prediction collection:
The multiple target prediction collection is voted, obtains integrated forecasting collection.
Optionally, second prediction module is additionally operable to:
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection.
Optionally, the model includes Random Forest model, arest neighbors model and logistic model,
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection, including:
Second training set is modeled using the Random Forest model, obtains first object forecast set;
Second training set is modeled using the arest neighbors model, obtains the second target prediction collection;
Second training set is modeled using the logistic model, obtains the 3rd target prediction collection.
Based on same inventive concept, third aspect present invention provides a kind of computer-readable recording medium, deposits thereon
Computer program is contained, the program realizes following steps when being executed by processor:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set is entered
Row prediction, obtains the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and based on the multiple
Target prediction collection carries out market prediction.
Based on same inventive concept, fourth aspect present invention provides a kind of computer equipment, including memory, processing
Device and storage are realized on a memory and the computer program that can run on a processor, during the computing device described program
Following steps:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set is entered
Row prediction, obtains the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and based on the multiple
Target prediction collection carries out market prediction.
The one or more technical schemes provided in the embodiment of the present invention, have at least the following technical effects or advantages:
A kind of processing method of market prediction data provided by the invention, the tentative prediction data set that will be obtained ahead of time first
As the first training set;Then first training set is classified using convolutional neural networks model, and to sorted
First training set is predicted, and obtains the first prediction result;First prediction result is added to described sorted again
In one training set, the second training set is obtained;Finally according to second training set, multiple targets are obtained respectively using bagging method
Forecast set, and market prediction is carried out based on the multiple target prediction collection.The above method of the present invention, on the one hand, using convolution
Neural network model is predicted again after classifying to tentative prediction data set, i.e., has carried out one-level to tentative prediction data set
Modeling, and convolution convolutional neural networks model is employed, therefore the accuracy of the first prediction result, the opposing party can be improved
Face, the first prediction result that convolutional neural networks model is obtained is added into sorted first training set, and is obtained
Second training set, then the second training set again, multiple target prediction collection is obtained using bagging method, i.e., to predictive data set respectively
Two modelings have been carried out, the defects of avoiding the prediction effect deficiency of single weak learner, have been obtained respectively using bagging method more
Individual target prediction collection, and multiple target prediction collection are merged, and then market prediction is carried out, target prediction collection can be caused more
Add closing to reality, so further improve the accuracy of prediction result.Method of the prior art is solved due to forecast model
Technical problem that is single and prediction result inaccuracy being present.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
There is the required accompanying drawing used in technology description to be briefly described, it should be apparent that, drawings in the following description are the present invention
Some embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can also basis
These accompanying drawings obtain other accompanying drawings.
Fig. 1 is the flow chart of the processing method of market prediction data in the embodiment of the present invention;
Fig. 2 is the structure chart of the processing unit of market prediction data in the embodiment of the present invention;
The structure chart of Fig. 3 Computer readable storage medium storing program for executing of the embodiment of the present invention;
Fig. 4 is the structure chart of Computer equipment of the embodiment of the present invention.
Embodiment
The invention provides a kind of processing method and processing device of market prediction data, to solve method of the prior art
The inaccurate technical problem of prediction result be present because forecast model is single.
Technical scheme in the embodiment of the present application, general thought are as follows:
A kind of processing method of market prediction data, first using the tentative prediction data set being obtained ahead of time as the first training
Collection;Then first training set is classified using convolutional neural networks model, and sorted first training set is entered
Row prediction, obtains the first prediction result;First prediction result is added into sorted first training set again, obtained
Obtain the second training set;Finally according to second training set, multiple target prediction collection are obtained respectively using bagging method, and be based on
The multiple target prediction collection carries out market prediction.
In the above method, on the one hand, enter again after being classified using convolutional neural networks model to tentative prediction data set
Row prediction, i.e., carried out one-level modeling to tentative prediction data set, and employs convolution convolutional neural networks model, therefore can
To improve the accuracy of the first prediction result, on the other hand, the first prediction result that convolutional neural networks model is obtained adds
To in sorted first training set, and the second training set is obtained, then the second training set again, distinguished using bagging method
Multiple target prediction collection are obtained, i.e., two modelings have been carried out to predictive data set, avoid the prediction effect of single weak learner
The defects of insufficient, obtain multiple target prediction collection respectively using bagging method, and multiple target prediction collection are merged, and then
Market prediction is carried out, target prediction collection more closing to reality can be caused, so further improve the accuracy of prediction result.Solution
The inaccurate technical problem of prediction result be present because forecast model is single in method of the prior art of having determined.
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
Part of the embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
A kind of processing method of market prediction data is present embodiments provided, refer to Fig. 1, methods described includes:
Step S101:Using the tentative prediction data set being obtained ahead of time as the first training set;
Step S102:First training set is classified using convolutional neural networks model, and to sorted
One training set is predicted, and obtains the first prediction result;
Step S103:First prediction result is added into sorted first training set, obtains the second instruction
Practice collection;
Step S104:According to second training set, multiple target prediction collection are obtained respectively using bagging method, and be based on
The multiple target prediction collection carries out market prediction.
Below, with reference to Fig. 1, a kind of processing method of the market prediction data provided the present embodiment is described in detail:
Step S101 is first carried out:Using the tentative prediction data set being obtained ahead of time as the first training set.
Specifically, the tentative prediction data set being obtained ahead of time can be the tentative prediction data obtained using existing method
Collection, such as according to the mode of statistics sales volume or the navigation patterns according to user etc..
Then step S102 is performed:First training set is classified using convolutional neural networks model, and to dividing
The first training set after class is predicted, and obtains the first prediction result.
Specifically, in the prior art, it is generally adopted by logistic method in two classification problems, but Shen of the present invention
Ask someone to find by long-term practice and research, easily realized although logistic method calculation cost is smaller, due to can
The problem of poor fitting and not high nicety of grading can easily be caused.Therefore, the embodiments of the invention provide a kind of convolutional neural networks
Method, because convolutional neural networks have the very strong feature extraction distinguishing ability by part to the overall situation, classification knot can be improved
The accuracy of fruit.Statistics and analysis by specific data, this is other to being drawn using logistic method and convolutional neural networks
Classification results verified that the result is that the error rate of logistic method is 24.172%, convolutional neural networks method
Error rate be 15.360%, therefore, using the present invention convolutional neural networks model first training set is classified,
The accuracy of classification can be improved, then sorted first training set is predicted, obtains the first prediction result, and by the
One prediction result retains.For example with the first-order model of convolutional neural networks method structure, the recurrence system of the first-order model is utilized
Number represents influence degree of each behavior to final result, can by taking user click frequency, purchase number and the first prediction result as an example
Using draw the number of clicks of user to the influence degree of final result as 1.0496465, influence of the purchase number to final result
Degree is 1.0226648, and the first prediction result is 6.408171 to the influence degree of final result, and above-mentioned numerical value is by calculating
Obtained different behavioural informations and the first prediction result influence the coefficient of size on final result, and numerical value is bigger, shows this feature
Second level result is influenceed bigger.Above-mentioned technical proposal can change according to whether prediction user buys come prediction markets, than straight
It is more accurate and there is practical significance to connect prediction markets, and predicts that the prediction result brought is not known compared to traditional quantitative analysis
Property, the above method of the invention can be by improving the prediction accuracy of first-order model, and so as to reach, to improve market prediction accurate
The technique effect of property so that market prediction anticipated that control.
Next step S103 is performed:First prediction result is added into sorted first training set,
Obtain the second training set.
Specifically, because abovementioned steps using convolutional neural networks model obtain the first prediction result, in order to enter one
Step improves the accuracy of prediction data, and the present invention adds the first prediction result obtained above to sorted first instruction
Practice and concentrate, the second training set is obtained, for follow-up data processing.
Finally perform step S104:According to second training set, multiple target predictions are obtained respectively using bagging method
Collection, and market prediction is carried out based on the multiple target prediction collection.
Specifically, it is described according to second training set, obtain multiple target prediction collection respectively using bagging method, and
Market prediction is carried out based on the multiple target prediction collection, including:
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection.
More specifically, the model includes Random Forest model, arest neighbors model and logistic model,
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection, including:
Second training set is modeled using the Random Forest model, obtains first object forecast set;
Second training set is modeled using the arest neighbors model, obtains the second target prediction collection;
Second training set is modeled using the logistic model, obtains the 3rd target prediction collection.
In specific implementation process, the embodiment of the present invention is entered using convolutional neural networks model to predictive data set first
Row one-level is modeled, and the result of one-level modeling is added into the first training set, so as to obtain the second training set, and by the second training set
As the source data of two level modeling, multiple target prediction collection are obtained using bagging method respectively in two level modeling, and based on described
Multiple target prediction collection carry out market prediction.Because two level modeling uses Bagging methods, single weak learner can be avoided
The shortcomings that prediction effect deficiency, and respectively from random forest, KNN (k-NearestNeighbor arest neighbors model), logic this
The base of a fruit such as returns at the single model, and random sampling is carried out to the second training set, and first object forecast set model_1 is respectively trained out, and two
Target prediction collection model_2 and the 3rd target prediction collection model_3, and above three target prediction collection is merged to obtain most
Whole forecast set, and then carry out market prediction.Specifically, according to the previous theorem of extensive error, for two classification problems, to appointing
A function anticipate all at least with probability δ so that
Wherein
R (f) represents two classification function f desired value,Two classification function f empiric risk is represented, d represents to assume sky
Between middle function number, N represent sample size, δ represent probable value.It can thus be appreciated that training error is smaller, then extensive error is smaller,
The effect of prediction will be better.And from empirical risk minimization principle, the model of empirical risk minimization is optimal models.It is false
If classification function is
f:Rn→{c1,c2}
The probability of misclassification is
P (Y ≠ f (X)=1-P (Y=f (X))
Wherein, RnRepresent n dimension real number vector spaces, c1,c2Two values of vector space are represented, Y represents input stochastic variable
Actual classification value, f (X) represents input random table amount X prediction classification value.The misclassification of the prediction result of so multiple models
Rate is
Wherein, K represents Number of Models, Nk(x) the stochastic variable space of k-th of model is represented.Can be with because voting
It is equivalent to empirical risk minimization.When data volume increases, the effect of empirical risk minimization can gradually improve.Due to this hair
It is bright to use two level fusion model, so that final forecast set is more accurate and reliable.
In order to further improve the accuracy of forecast set, according to second training set, obtained respectively using bagging method
After obtaining multiple target prediction collection, methods described also includes:
The multiple target prediction collection is voted, obtains integrated forecasting collection.
In specific implementation process, present invention applicant is had found by largely testing with after investigation and comparison, using dress
Multiple target prediction collection that bag method obtains, still and actual conditions have deviation, and therefore, the present invention is to the multiple target prediction
Collection is voted, and obtains integrated forecasting collection.Because second level model obtains multiple target prediction collection using bagging method,
Multiple models are employed to be predicted the second training set, therefore each model can obtain a result, 0 or 1, wherein 0 table
Showing to buy, and 1 represents to buy, and this multiple result can be carried out into number statistics by voting, if result is 1
Number is more than the number that result is 0, then it is assumed that final prediction result is 1, and on the contrary then prediction result is 0.Pass through the side of the present invention
Method, when being trained in the second level, as a result have preferably explanatory, for example, influence of the first order prediction result to the second level
Coefficient is 6.408171, and the influence coefficient of user's hits this behaviors to the second level is 1.0496465, then the first prediction
As a result it is 6 times of user's hits this behaviors to the influence degree of final result, that is, the user of prediction purchase for the first time
Final prediction result improves 6 times for the probability of purchase, thus data is more pressed close to model truth, and the present invention uses two
Level Fusion Model is handled market prediction data, and preliminary treatment, structure are being carried out to data using convolutional neural networks method
After building first-order model, on this basis, then using bagging method data are further processed, with the prediction knot of first-order model
Fruit is added in second-level model, the problem of avoiding single-stage or bad single modelling effect, so improve the standard of market prediction
True property.
Based on the inventive concept same with embodiment one, present invention also offers a kind of processing method of market prediction data
Corresponding device, referring specifically to embodiment two.
Embodiment two
A kind of processing unit of market prediction data is present embodiments provided, refers to Fig. 2, described device includes:
First obtains module 201, for the tentative prediction data set that will be obtained ahead of time as the first training set;
First prediction module 202, for being classified using convolutional neural networks model to first training set, and it is right
Sorted first training set is predicted, and obtains the first prediction result;
Second obtains module 203, for first prediction result to be added into sorted first training set,
Obtain the second training set;
Second prediction module 204, for according to second training set, it is pre- to obtain multiple targets respectively using bagging method
Collection is surveyed, and market prediction is carried out based on the multiple target prediction collection.
In the processing unit of market prediction data provided in an embodiment of the present invention, described device also includes processing module, uses
In after multiple target prediction collection according to second training set, are obtained respectively using bagging method:
The multiple target prediction collection is voted, obtains integrated forecasting collection.
In the processing unit of market prediction data provided in an embodiment of the present invention, second prediction module is additionally operable to:
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection.
In the processing unit of market prediction data provided in an embodiment of the present invention, the model include Random Forest model,
Arest neighbors model and logistic model,
A variety of models are respectively adopted to be modeled second training set, obtain target prediction corresponding with each model
Collection, including:
Second training set is modeled using the Random Forest model, obtains first object forecast set;
Second training set is modeled using the arest neighbors model, obtains the second target prediction collection;
Second training set is modeled using the logistic model, obtains the 3rd target prediction collection.
The various change mode and instantiation of the processing method of market prediction data in embodiment one are equally applicable to
The device of the present embodiment, by the detailed description of the foregoing processing method to market prediction data, those skilled in the art can be with
The device being apparent from the present embodiment, thus it is succinct for specification, it will not be described in detail herein.
Based on the inventive concept same with embodiment one, present invention also offers a kind of processing method of market prediction data
Corresponding computer-readable recording medium, referring specifically to embodiment three.
Embodiment three
A kind of computer-readable recording medium 300 is present embodiments provided, is stored thereon with computer program 311, the journey
Following steps are realized when sequence is executed by processor:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set is entered
Row prediction, obtains the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and based on the multiple
Target prediction collection carries out market prediction.
The various change mode and instantiation of the processing method of market prediction data in embodiment one are equally applicable to
The computer-readable recording medium of the present embodiment, pass through the detailed description of the foregoing processing method to market prediction data, ability
Field technique personnel are clear that the computer-readable recording medium in the present embodiment, thus it is succinct for specification,
It will not be described in detail herein.
Based on the inventive concept same with embodiment one, present invention also offers a kind of processing method of market prediction data
Corresponding computer-readable recording medium, referring specifically to example IV.
Example IV
A kind of computer equipment is present embodiments provided, including memory 401, processor 402 and storage are on a memory
And the computer program 403 that can be run on a processor, following steps are realized during the computing device described program:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set is entered
Row prediction, obtains the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and based on the multiple
Target prediction collection carries out market prediction.
For convenience of description, Fig. 4 illustrate only the part related to the embodiment of the present invention, and particular technique details does not disclose
, it refer to present invention method part.Wherein, memory 401 can be used for storage software program and module, processor
402 perform the software program and module that are stored in memory 401 by running, should so as to perform the various functions of mobile terminal
With and data processing.
Memory 401 can mainly include storing program area and storage data field, wherein, storing program area can store operation system
Application program needed for system, at least one function etc.;Storage data field can store uses what is created according to computer equipment
Data etc..The control centre of the mobile communication terminal of processor 402, utilize various interfaces and the whole mobile communication terminal of connection
Various pieces, by running or performing the software program and/or module that are stored in memory 501, and call and be stored in
Data in memory 401, the various functions and processing data of mobile phone are performed, it is overall so as to be carried out to mobile phone
Monitoring.
The various change mode and instantiation of the processing method of market prediction data in embodiment one are equally applicable to
The computer equipment of the present embodiment, pass through the detailed description of the foregoing processing method to market prediction data, people in the art
Member is clear that computer equipment in the present embodiment, thus it is succinct for specification, it will not be described in detail herein.
The one or more technical schemes provided in the embodiment of the present invention, have at least the following technical effects or advantages:
A kind of processing method of market prediction data provided by the invention, the tentative prediction data set that will be obtained ahead of time first
As the first training set;Then first training set is classified using convolutional neural networks model, and to sorted
First training set is predicted, and obtains the first prediction result;First prediction result is added to described sorted again
In one training set, the second training set is obtained;Finally according to second training set, multiple targets are obtained respectively using bagging method
Forecast set, and market prediction is carried out based on the multiple target prediction collection.The above method of the present invention, on the one hand, using convolution
Neural network model is predicted again after classifying to tentative prediction data set, i.e., has carried out one-level to tentative prediction data set
Modeling, and convolution convolutional neural networks model is employed, therefore the accuracy of the first prediction result, the opposing party can be improved
Face, the first prediction result that convolutional neural networks model is obtained is added into sorted first training set, and is obtained
Second training set, then the second training set again, multiple target prediction collection is obtained using bagging method, i.e., to predictive data set respectively
Two modelings have been carried out, the defects of avoiding the prediction effect deficiency of single weak learner, have been obtained respectively using bagging method more
Individual target prediction collection, and multiple target prediction collection are merged, and then market prediction is carried out, target prediction collection can be caused more
Add closing to reality, so further improve the accuracy of prediction result.Method of the prior art is solved due to forecast model
Technical problem that is single and prediction result inaccuracy being present.
It should be understood by those skilled in the art that, embodiments of the invention can be provided as method, system or computer program
Product.Therefore, the present invention can use the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware
Apply the form of example.Moreover, the present invention can use the computer for wherein including computer usable program code in one or more
The computer program production that usable storage medium is implemented on (including but is not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of product.
The present invention is the flow with reference to method according to embodiments of the present invention, equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that can be by every first-class in computer program instructions implementation process figure and/or block diagram
Journey and/or the flow in square frame and flow chart and/or block diagram and/or the combination of square frame.These computer programs can be provided
The processors of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce
A raw machine so that produced by the instruction of computer or the computing device of other programmable data processing devices for real
The device for the function of being specified in present one flow of flow chart or one square frame of multiple flows and/or block diagram or multiple square frames.
These computer program instructions, which may be alternatively stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that the instruction being stored in the computer-readable memory, which produces, to be included referring to
Make the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one square frame of block diagram or
The function of being specified in multiple square frames.
These computer program instructions can be also loaded into computer or other programmable data processing devices so that counted
Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented processing, so as in computer or
The instruction performed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in individual square frame or multiple square frames.
Although preferred embodiments of the present invention have been described, but those skilled in the art once know basic creation
Property concept, then can make other change and modification to these embodiments.So appended claims be intended to be construed to include it is excellent
Select embodiment and fall into having altered and changing for the scope of the invention.
Obviously, those skilled in the art can carry out various changes and modification without departing from this hair to the embodiment of the present invention
The spirit and scope of bright embodiment.So, if these modifications and variations of the embodiment of the present invention belong to the claims in the present invention
And its within the scope of equivalent technologies, then the present invention is also intended to comprising including these changes and modification.
Claims (10)
1. a kind of processing method of market prediction data, it is characterised in that methods described includes:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set carried out pre-
Survey, obtain the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and be based on the multiple target
Forecast set carries out market prediction.
2. the method as described in claim 1, it is characterised in that according to second training set, distinguished using bagging method
After obtaining multiple target prediction collection, methods described also includes:
The multiple target prediction collection is voted, obtains integrated forecasting collection.
3. the method as described in claim 1, it is characterised in that it is described according to second training set, using bagging method point
Multiple target prediction collection are not obtained, and market prediction is carried out based on the multiple target prediction collection, including:
A variety of models are respectively adopted to be modeled second training set, obtain target prediction collection corresponding with each model.
4. method as claimed in claim 3, it is characterised in that the model include Random Forest model, arest neighbors model and
Logistic model,
A variety of models are respectively adopted to be modeled second training set, obtain target prediction collection corresponding with each model,
Including:
Second training set is modeled using the Random Forest model, obtains first object forecast set;
Second training set is modeled using the arest neighbors model, obtains the second target prediction collection;
Second training set is modeled using the logistic model, obtains the 3rd target prediction collection.
5. a kind of processing unit of market prediction data, it is characterised in that described device includes:
First obtains module, for the tentative prediction data set that will be obtained ahead of time as the first training set;
First prediction module, for being classified to first training set using convolutional neural networks model, and to classification after
The first training set be predicted, obtain the first prediction result;
Second obtains module, for first prediction result to be added into sorted first training set, obtains the
Two training sets;
Second prediction module, for according to second training set, obtaining multiple target prediction collection respectively using bagging method, and
Market prediction is carried out based on the multiple target prediction collection.
6. device as claimed in claim 5, it is characterised in that described device also includes processing module, for according to
Second training set, after obtaining multiple target prediction collection respectively using bagging method:
The multiple target prediction collection is voted, obtains integrated forecasting collection.
7. device as claimed in claim 5, it is characterised in that second prediction module is additionally operable to:
A variety of models are respectively adopted to be modeled second training set, obtain target prediction collection corresponding with each model.
8. method as claimed in claim 7, it is characterised in that the model include Random Forest model, arest neighbors model and
Logistic model,
A variety of models are respectively adopted to be modeled second training set, obtain target prediction collection corresponding with each model,
Including:
Second training set is modeled using the Random Forest model, obtains first object forecast set;
Second training set is modeled using the arest neighbors model, obtains the second target prediction collection;
Second training set is modeled using the logistic model, obtains the 3rd target prediction collection.
9. a kind of computer-readable recording medium, is stored thereon with computer program, it is characterised in that the program is held by processor
Following steps are realized during row:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set carried out pre-
Survey, obtain the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and be based on the multiple target
Forecast set carries out market prediction.
10. a kind of computer equipment, including memory, processor and storage are on a memory and the meter that can run on a processor
Calculation machine program, it is characterised in that realize following steps during the computing device described program:
Using the tentative prediction data set being obtained ahead of time as the first training set;
First training set is classified using convolutional neural networks model, and sorted first training set carried out pre-
Survey, obtain the first prediction result;
First prediction result is added into sorted first training set, obtains the second training set;
According to second training set, multiple target prediction collection are obtained respectively using bagging method, and be based on the multiple target
Forecast set carries out market prediction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710735073.9A CN107633421A (en) | 2017-08-24 | 2017-08-24 | A kind of processing method and processing device of market prediction data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710735073.9A CN107633421A (en) | 2017-08-24 | 2017-08-24 | A kind of processing method and processing device of market prediction data |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107633421A true CN107633421A (en) | 2018-01-26 |
Family
ID=61101381
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710735073.9A Pending CN107633421A (en) | 2017-08-24 | 2017-08-24 | A kind of processing method and processing device of market prediction data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107633421A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109064203A (en) * | 2018-06-12 | 2018-12-21 | 佛山欧神诺陶瓷有限公司 | A kind of BI data statistics method for early warning |
CN110046636A (en) * | 2018-12-11 | 2019-07-23 | 阿里巴巴集团控股有限公司 | Prediction technique of classifying and device, prediction model training method and device |
CN112766514A (en) * | 2021-01-22 | 2021-05-07 | 支付宝(杭州)信息技术有限公司 | Method, system and device for joint training of machine learning model |
CN113837323A (en) * | 2021-11-08 | 2021-12-24 | 中国联合网络通信集团有限公司 | Satisfaction prediction model training method and device, electronic equipment and storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012121747A1 (en) * | 2011-03-04 | 2012-09-13 | Ultratick, Inc. | Predicting the performance of a financial instrument |
CN102890803A (en) * | 2011-07-21 | 2013-01-23 | 阿里巴巴集团控股有限公司 | Method and device for determining abnormal transaction process of electronic commodity |
CN105427129A (en) * | 2015-11-12 | 2016-03-23 | 腾讯科技(深圳)有限公司 | Information delivery method and system |
CN106845681A (en) * | 2016-12-21 | 2017-06-13 | 广东工业大学 | A kind of stock trend forecasting method of application depth learning technology |
-
2017
- 2017-08-24 CN CN201710735073.9A patent/CN107633421A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012121747A1 (en) * | 2011-03-04 | 2012-09-13 | Ultratick, Inc. | Predicting the performance of a financial instrument |
CN102890803A (en) * | 2011-07-21 | 2013-01-23 | 阿里巴巴集团控股有限公司 | Method and device for determining abnormal transaction process of electronic commodity |
CN105427129A (en) * | 2015-11-12 | 2016-03-23 | 腾讯科技(深圳)有限公司 | Information delivery method and system |
CN106845681A (en) * | 2016-12-21 | 2017-06-13 | 广东工业大学 | A kind of stock trend forecasting method of application depth learning technology |
Non-Patent Citations (1)
Title |
---|
WILL_DUAN: "集成学习总结 & Stacking方法详解", 《HTTPS://BLOG.CSDN.NET/WILLDUAN1/ARTICLE/DETAILS/73618677》 * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109064203A (en) * | 2018-06-12 | 2018-12-21 | 佛山欧神诺陶瓷有限公司 | A kind of BI data statistics method for early warning |
CN110046636A (en) * | 2018-12-11 | 2019-07-23 | 阿里巴巴集团控股有限公司 | Prediction technique of classifying and device, prediction model training method and device |
CN112766514A (en) * | 2021-01-22 | 2021-05-07 | 支付宝(杭州)信息技术有限公司 | Method, system and device for joint training of machine learning model |
CN113837323A (en) * | 2021-11-08 | 2021-12-24 | 中国联合网络通信集团有限公司 | Satisfaction prediction model training method and device, electronic equipment and storage medium |
CN113837323B (en) * | 2021-11-08 | 2023-07-04 | 中国联合网络通信集团有限公司 | Training method and device of satisfaction prediction model, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN106485562B (en) | Commodity information recommendation method and system based on user historical behaviors | |
CN107220217A (en) | Characteristic coefficient training method and device that logic-based is returned | |
CN109919684A (en) | For generating method, electronic equipment and the computer readable storage medium of information prediction model | |
CN108734338A (en) | Credit risk forecast method and device based on LSTM models | |
CN108256907A (en) | A kind of construction method and computing device of customer grouping model | |
CN104798043B (en) | A kind of data processing method and computer system | |
CN107633421A (en) | A kind of processing method and processing device of market prediction data | |
CN109359812A (en) | Finance product recommended method, server and computer readable storage medium | |
CN106530010B (en) | The collaborative filtering method and device of time of fusion factor | |
CN109446430A (en) | Method, apparatus, computer equipment and the readable storage medium storing program for executing of Products Show | |
CN108256568A (en) | A kind of plant species identification method and device | |
CN108960719A (en) | Selection method and apparatus and computer readable storage medium | |
CN108898476A (en) | A kind of loan customer credit-graded approach and device | |
CN107230108A (en) | The processing method and processing device of business datum | |
CN107423613A (en) | The method, apparatus and server of device-fingerprint are determined according to similarity | |
CN109657890A (en) | A kind of risk for fraud of transferring accounts determines method and device | |
CN107818491A (en) | Electronic installation, Products Show method and storage medium based on user's Internet data | |
CN109816509A (en) | Generation method, terminal device and the medium of scorecard model | |
CN105468628B (en) | A kind of sort method and device | |
CN110210024A (en) | A kind of information processing method, device and storage medium | |
CN107545038A (en) | A kind of file classification method and equipment | |
CN107368526A (en) | A kind of data processing method and device | |
CN112036483B (en) | AutoML-based object prediction classification method, device, computer equipment and storage medium | |
CN107239532A (en) | Data digging method and device | |
CN109598285A (en) | A kind of processing method of model, device and equipment |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180126 |
|
RJ01 | Rejection of invention patent application after publication |