CN106600312A - Price prediction method and system for wine buying of consumer - Google Patents
Price prediction method and system for wine buying of consumer Download PDFInfo
- Publication number
- CN106600312A CN106600312A CN201611041660.XA CN201611041660A CN106600312A CN 106600312 A CN106600312 A CN 106600312A CN 201611041660 A CN201611041660 A CN 201611041660A CN 106600312 A CN106600312 A CN 106600312A
- Authority
- CN
- China
- Prior art keywords
- factor
- influence
- consumer
- classification
- wine
- 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
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
Abstract
The embodiment of the invention provides a price prediction method and a system for wine buying of a consumer. The method comprises steps: influence elements corresponding to an acquired consumer sample set are extracted, wherein each influence element at least comprises an influence factor; the weight corresponding to each influence factor is calculated, and a target influence factor is acquired according to the weight; a two-category v double-support vector machine model is built according to the target influence factor and the wine price, a multi-category model is built according to the two-category v double-support vector machine model, and the multi-category model is trained by using the consumer sample set; and the multi-category model is used for predicting a to-be-predicted sample. The system is used for executing the method. according to the embodiment of the invention, through building the two-category v double-support vector machine model based on the target influence factor and the wine price and building the multi-category model based on the two-category v double-support vector machine model, the accuracy for predicting the price of the wine bought by the consumer can be improved.
Description
Technical field
The present embodiments relate to data mining technology field, more particularly to a kind of price prediction side of consumer purchase wine
Method and system.
Background technology
With the improvement of people's living standards, the species and price of wine on the market tend to variation, research consumption
The consuming behavior of person formulates marketing strategy to manufacturing enterprise, increases the aspects such as business income and has important practical significance.
By taking wine as an example, wine is increasingly subject to people's attention as a kind of beverage of healthy fashion again, and which disappears
Expense is also increasingly popularized.The consumption preferences and consuming behavior trend of consumer can largely affect walking for Wine Market
To, and then affect the direction of China's wine industry future development.Understand wine consumer in depth to be studied, can both enter
One step is improved and expands consumer behavior research theory, can more make the marketing activity of Wine Enterprises have more specific aim.It is existing
Used in technology, two classify the double support vector machine training patterns of v to be predicted the behavior of consumer, but the method prediction
Degree of accuracy it is not high.Therefore, how to improve to the degree of accuracy of consumer's price of purchase prediction is problem nowadays urgently to be resolved hurrily.
The content of the invention
For the problem that prior art is present, the embodiment of the present invention provides the price Forecasting Methodology that a kind of consumer buys wine
And system.
On the one hand, the embodiment of the present invention provides the price Forecasting Methodology that a kind of consumer buys wine, including：
The corresponding influence factor of consumer's sample set to getting extracts, and the influence factor includes：Quality because
Element, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor, each influence factor at least including one affect because
Son；
The corresponding weight of each factor of influence is calculated, and according to the Weight Acquisition object effects factor；
The double supporting vector machine models of two classification v are set up according to the price of the object effects factor and wine, according to described two
The double supporting vector machine models of classification v set up many disaggregated models, and many disaggregated models are entered using consumer's sample set
Row training；
Forecast sample is treated using many disaggregated models to be predicted.
On the other hand, the embodiment of the present invention provides the price prognoses system that a kind of consumer buys wine, including：
Extraction module, extracts for the corresponding influence factor of consumer's sample set to getting, the impact because
Element includes：Qualitative factor, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor, each influence factor is at least
Including a factor of influence；
Computing module, for calculating the corresponding weight of each factor of influence, and according to the Weight Acquisition object effects
The factor；
Model building module, for setting up the double supporting vectors of two classification v according to the price of the object effects factor and wine
Machine model, sets up many disaggregated models according to the double supporting vector machine models of described two classification v, and utilizes consumer's sample set pair
Many disaggregated models are trained；
Model prediction module, is predicted for treating forecast sample using many disaggregated models.
The price Forecasting Methodology and system of a kind of consumer's purchase wine provided in an embodiment of the present invention, by according to target shadow
The price for ringing the factor and wine sets up the double supporting vector machine models of two classification v, sets up many according to the double supporting vector machine models of two classification v
Disaggregated model, improves the accuracy of the price prediction that wine is bought to consumer.
Description of the drawings
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
Accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are these
Some bright embodiments, for those of ordinary skill in the art, on the premise of not paying creative work, can be with root
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is the price Forecasting Methodology schematic flow sheet that a kind of consumer provided in an embodiment of the present invention buys wine；
Fig. 2 is the double supporting vector machine model classification schematic diagrams of provided in an embodiment of the present invention two classification v；
Fig. 3 is many disaggregated model training schematic diagrams provided in an embodiment of the present invention；
Fig. 4 is Lasso algorithms schematic diagram calculation provided in an embodiment of the present invention；
Fig. 5 buys the price Forecasting Methodology schematic flow sheet of wine for a kind of consumer that another embodiment of the present invention is provided；
Fig. 6 predicts schematic flow sheet for many disaggregated models provided in an embodiment of the present invention；
Fig. 7 is the price prognoses system structural representation that a kind of consumer provided in an embodiment of the present invention buys wine；
Fig. 8 buys the price prognoses system structural representation of wine for a kind of consumer that another embodiment of the present invention is provided.
Specific embodiment
To make purpose, technical scheme and the 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
The a part of embodiment of the present invention, rather than the embodiment of whole.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.
Fig. 1 is the price Forecasting Methodology schematic flow sheet that a kind of consumer provided in an embodiment of the present invention buys wine, such as Fig. 1
It is shown, methods described, including：
Step 101：The corresponding influence factor of consumer's sample set to getting extracts, influence factor's bag
Include：Qualitative factor, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor, each influence factor at least include
One factor of influence；
Specifically, consumer's sample set is obtained, its acquisition modes can be the mode of questionnaire by inquiry, but be not limited to
Aforesaid way, the corresponding influence factor of consumer's sample set to getting are extracted, and influence factor includes qualitative factor, enterprise
Industry Marketing Factors, motive in purchasing factor and individual characteristicss factor etc..Wherein affect wine qualitative factor have color, fragrance, mouthfeel,
Whether win a prize, the place of production, the factor of influence such as time and packaging；Enterprise marketing factor includes：Advertising input, advertising campaign, sales field
The factors of influence such as the service skill of geographical position and shop-assistant；Motive in purchasing factor includes：Give a present, entertain, meetting and oneself
The factors of influence such as drink；Individual characteristicss factor includes：The income of consumer, educational background, occupation, age, sex and marriage situation etc.
Factor of influence.
Step 102：The corresponding weight of each factor of influence is calculated, and according to the Weight Acquisition object effects factor；
Specifically, calculate the corresponding weight of above-mentioned each factor of influence, it is to be understood that the bigger factor of influence of weight
Which is bigger to the influence degree for predicting the outcome, and the factor of influence is more important, according to the Weight Acquisition object effects factor.
Step 103：Double supporting vector machine model (the v- of two classification v are set up according to the price of the object effects factor and wine
TSVM), many disaggregated models are set up according to the double supporting vector machine models of described two classification v, and utilizes consumer's sample set pair
Many disaggregated models (v-TDAGSVM) are trained；
Specifically, Fig. 2 is the double supporting vector machine models classification schematic diagrams of provided in an embodiment of the present invention two classification v, such as Fig. 2
It is shown, according to the object effects factor and the price of wine, the double supporting vector machine models of two classification v, the wherein price of wine are set up two-by-two
It is divided into 0-100 units, 101-200 units, 201-300 units and more than 300 yuan, it is to be understood that the price of wine can be according to reality
Border situation is divided, and is not specifically limited herein.If the price of wine there are K classes, K (K-1)/2 two classification v can be set up double
Supporting vector machine model；Wherein linear model such as formula (1) and formula (2) are shown：
Wherein, XiAnd XjConsumer's sample set of consumer's composition of the price of i-th kind and jth kind wine is bought respectively；WithFor complete 1 column vector；νi∈ (0,1) and νj∈ (0, it is 1) parameter given in advance, for controlling two classification
The fraction of the number and wrong point of rate of the supporting vector in the double supporting vector machine models of v；ξijAnd ξjiBe slack variable composition to
Amount；li,ljFor i-th kind of consumption and consumer's number of jth kind wine.
In formula (1), the requirement that object function Section 1 is represented minimizes each sample point of the i-th class to its corresponding hyperplane
Distance quadratic sum, its meaning is, it is desirable to the sample point of same type, this refers to the i-th class sample point and leans against as far as possible
Together；Section 2 represents ρ to be maximizediValue；Section 3 represents to the i-th class hyperplane mistake minute rate minimized by jth class o'clock.
Constraints causes the point ρ as far as possible of jth classiDistance with away from the i class hyperplane.Similar, formula (2) also has upper
Theory significance is stated, here is omitted.
It is used for seeking following two uneven hyperplane by above-mentioned model：
fij=(wij·x)+bij=0 and fji=(wji·x)+bji=0 (3)
Wherein, wij,wji∈RnIt is the normal vector of hyperplane in n dimension theorem in Euclid space；bij,bji∈ R are biasings.
When needing to test new sample point (consumer) x, the distance that need to calculate it apart from the two hyperplane is big
It is little：
With
If dij＜ dji, that is to say, that the distance of test sample o'clock to the i-th class hyperplane is super flat to jth class less than it
The distance in face, then differentiate that this new sample point belongs to the i-th class, that is, judge that the consumer can consume the wine of the i-th class price；
Vice versa.
According to the double supporting vector machine model of K (K-1)/2 two of above-mentioned foundation classification v, selected by directed acyclic graph and
Many disaggregated models are rebuild, and many disaggregated models is trained using the consumer's sample set for getting.
Fig. 3 is many disaggregated model training schematic diagrams provided in an embodiment of the present invention, as shown in Figure 3, it is assumed that have classification 1, class
Other 2 and the wine of 3 these three prices of classification.First, double of two classification v are set up with the known sample for belonging to classification 1 and classification 2
Vector machine model is held, hyperplane 12 and hyperplane 21 constructed by formula (3) is obtained, judges that test sample belongs to using formula (4)
In classification 1 or classification 2；Same process, builds the double supports of two classification v with the known sample for belonging to classification 2 and classification 3 respectively
Vector machine model, obtains hyperplane 23 and hyperplane 32 that formula (3) is set up, judges that test sample belongs to using formula (4)
Classification 2 or 3；To classification 3 and classification 1 in the same manner.Such one meets the double supporting vector machine models of 3 two classification v of structure together.
Step 104：Forecast sample is treated using many disaggregated models to be predicted.
Specifically, it is predicted using training many disaggregated models and can treat forecast sample.
The embodiment of the present invention is by setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of two classification v, improves the price prediction to consumer's purchase wine
Accuracy.
It is on the basis of above-described embodiment, described that many classification moulds are set up according to the double supporting vector machine models of described two classification v
Type to many disaggregated models, and many disaggregated models are trained using consumer's sample set, also include：
The optimum described many disaggregated models of model parameter are obtained by cross-validation method.
Specifically, because many disaggregated models for building depend on previously given parameter v, particularly, for non-thread disposition
Shape, also relies on nuclear parameter r.It is in the process using cross validation method choosing optimized parameter, accurate with the classification for ensureing model
True property.The process approach is as follows：Each class sample in all of consumer's sample set is all divided into into 5 parts, each apoplexy due to endogenous wind
4 parts therein are extracted as training, it is remaining for testing.This process is repeated 5 times, the model parameter chosen in 5 times is extremely
Rare 2 is different, selects the corresponding model parameter of accuracy rate highest, so that many disaggregated models are optimum.
The embodiment of the present invention is trained to model so as to obtain optimum many disaggregated models by cross-validation method, is improved
The accuracy of prediction.
It is on the basis of above-described embodiment, described to calculate the corresponding weight of each factor of influence, including：
The corresponding weight of each factor of influence is calculated according to Lasso algorithms.
Specifically, Fig. 4 is Lasso algorithms schematic diagram calculation provided in an embodiment of the present invention, as shown in figure 4, X1…Xk1For
The corresponding factor of influence of product quality factor, Xk1+1…Xk2For the corresponding factor of influence of enterprise marketing factor, Xk2+1…Xk3For purchase
Buy the corresponding factor of influence of motivational factor, Xkk+1…XNFor the corresponding factor of influence of individual characteristicss factor.Lasso algorithms are a kind of
Shrinkage estimation.It obtains the model of a more refine by constructing penalty function so that it compresses some coefficients, while setting
Fixed some coefficients are zero.Therefore the advantage of subset contraction is remained, is that a kind of process has the biased estimation of multi-collinearity data.
The basic thought of Lasso be regression coefficient absolute value sum less than a constant constraints under, make residual sum of squares (RSS)
Minimize such that it is able to produce some regression coefficients exactly equal to 0, obtain the model that can be explained.Therefore can pass through
Lasso algorithms calculate the corresponding weight of each factor of influence.For the corresponding formula of Lasso algorithms, such as shown in formula (5)：
Wherein,XjTo affect the factor of influence of consumer spending；The vector of the price composition of wine, y are bought for consumeriFor the price of the wine of consumer's purchase；βjTo need to ask
The weight factor for obtaining.
Solution is optimized by the model, you can try to achieve βj, that is, affect the weight of the consumption factor.
The embodiment of the present invention calculates the corresponding weight of each factor of influence by Lasso algorithms so that each factor of influence
Feature have more intuitive and interpretability.
On the basis of above-described embodiment, double of two classification v are set up in the price according to the factor of influence and wine
Before holding vector machine model, also include：
The object effects factor of the weight more than predetermined threshold value is obtained, wherein the predetermined threshold value is more than 0.
Specifically, before the double supporting vector machine models of two classification v are set up according to the price of the factor of influence and wine, from
In the object effects factor for getting obtain weight more than predetermined threshold value the object effects factor, wherein predetermined threshold value be more than 0, and
Can be adjusted according to practical situation.
The embodiment of the present invention by obtain weight more than predetermined threshold value the object effects factor, by weight minor impact because
Son removes, only the remaining weight large effect factor, simplifies the double supporting vector machine models of two classification v of foundation.
On the basis of the various embodiments described above, described v pair of supporting vector machine model of two classification is at least three.
Specifically, due to needing to set up many disaggregated models according to the double supporting vector machine models of two classification v set up, so two
The double supporting vector machine models of classification v are at least three.
The embodiment of the present invention is by setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of two classification v, improves the price prediction to consumer's purchase wine
Accuracy.
Fig. 5 buys the price Forecasting Methodology schematic flow sheet of wine for a kind of consumer that another embodiment of the present invention is provided,
As shown in figure 5, methods described overall flow is as follows：
By taking consumer's purchase wine price prediction as an example：
Step 501：Process of data preprocessing；By inquiry questionnaire determine affect consumer purchase wine each impact because
Element, including：The qualitative factor of wine, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor etc..Wherein affect
The qualitative factor of wine include the color of wine, fragrance, mouthfeel, whether win a prize, the place of production, time and packaging etc.；Enterprise
Marketing Factors include service skill of advertising input, advertising campaign, sales field geographical position and shop-assistant etc.；Motive in purchasing factor
Including give a present, entertain, meet and from drink etc.；Individual characteristicss factor includes the income of consumer, educational background, occupation, age, sex
And marriage situation etc.；By the weight of each factor of influence of Lasso algorithms calculating impact wine consuming behavior, and according to
The corresponding object effects factor of Weight Acquisition.
Step 502：Set-up procedure；Target shadow of the weight more than predetermined threshold value is obtained from the object effects factor for getting
The factor is rung, wherein predetermined threshold value is more than 0, and can be adjusted according to practical situation；Consumer's sample set of acquisition is divided into
Training data sample and test data sample.
Step 503：Training process；The double supporting vectors of two classification v are set up according to the price of the object effects factor and wine
Machine model, and many disaggregated models are set up by directed acyclic graph (DAGSVM algorithms) according to the double supporting vector machine models of two classification v,
It is corresponding most using the model parameter that cross-validation method obtains many disaggregated models by training data sample and test data sample
Excellent parameter, and multimode shape parameter is trained.
Step 504：Prediction process；The data of unknown consumer are predicted by many disaggregated models for training.Fig. 6
Schematic flow sheet is predicted for many disaggregated models provided in an embodiment of the present invention, as shown in Figure 6：Assume that the price of wine is divided into 4
Class, 0-100 units are 1 class；101-200 units are 2 classes；201-300 units are 3 classes；More than 300 yuan is 4 classes；Therefore 6 can be set up
The double supporting vector machine models (v-TSVM) of two classification v, 2 v-TSVM of respectively 1 vs；1 vs 3 v-TSVM；1 vs 4 v-
TSVM；2 vs 3 v-TSVM；2 vs 4 v-TSVM；3 vs 4 v-TSVM；Using more than, 6 v-TSVM set up many classification moulds
Type, wherein, the position of each v-TSVM in Fig. 4 can exchange.It is after having a sample to be predicted to be input into many disaggregated models, first
It is introduced into 1 vs, 4 v-TSVM；Judge that the classification of the sample to be predicted flows to 2 vs if being not belonging to 1 class if 1 vs, 4 v-TSVM
4 v-TSVM, if 2 vs, 4 v-TSVM judge that the sample to be predicted is not belonging to 2 classes, flow to 3 vs, 4 v-TSVM, 3 vs 4
V-TSVM can export the classification of the sample to be predicted and be belonging to 3 classes and still fall within 4 classes, and above prediction process is an act
Example, depending on its concrete flow direction is needed according to practical situation, the embodiment of the present invention is not especially limited to this.
The embodiment of the present invention is by setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of two classification v, improves the price prediction to consumer's purchase wine
Accuracy.
Fig. 7 is the price prognoses system structural representation that a kind of consumer provided in an embodiment of the present invention buys wine, such as Fig. 7
It is described, the system, including 701 computing module 702 of extraction module, model building module 703 and model prediction module 704, its
In：
Extraction module 701 is extracted for the corresponding influence factor of consumer's sample set to getting, the impact
Factor includes：Qualitative factor, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor, each influence factor is extremely
Include a factor of influence less；Computing module 702 is used to calculate the corresponding weight of each factor of influence, and according to the weight
Obtain the object effects factor；Model building module 703 is for setting up two classification v according to the price of the object effects factor and wine
Double supporting vector machine models, set up many disaggregated models according to the double supporting vector machine models of described two classification v, and utilize the consumption
Person's sample set is trained to many disaggregated models；Model prediction module 704 is for treating pre- using many disaggregated models
Test sample is originally predicted.
Specifically, extraction module 701 obtains consumer's sample set, and its acquisition modes can be the side of questionnaire by inquiry
Formula, but aforesaid way is not limited to, the corresponding influence factor of consumer's sample set to getting extracts, influence factor's bag
Include qualitative factor, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor etc..The qualitative factor of wine is wherein affected to have face
Color, fragrance, mouthfeel, whether win a prize, the place of production, the factor of influence such as time and packaging；Enterprise marketing factor includes：Advertising input,
Service skill of advertising campaign, sales field geographical position and shop-assistant etc.；Motive in purchasing factor includes：Give a present, entertain, meet
And from drink etc.；Individual characteristicss factor includes：The income of consumer, educational background, occupation, age, sex and marriage situation etc..Meter
Calculate module 702 calculate the corresponding weight of above-mentioned each factor of influence, it is to be understood that the bigger factor of influence of weight its to pre-
The influence degree for surveying result is bigger, and the factor of influence is more important, according to the Weight Acquisition object effects factor.Model building module
703 set up the double supporting vector machine models of two classification v according to the price of the object effects factor and wine, according to two classification v pairs
Hold vector machine model and set up many disaggregated models, and many disaggregated models are trained using consumer's sample set；Mould
Type prediction module 704 is treated forecast sample using many disaggregated models and is predicted.
A kind of price prognoses system of consumer's purchase wine that the present invention is provided is used to perform said method, its specific reality
Apply mode consistent with the embodiment of method, here is omitted.
The embodiment of the present invention is by setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of two classification v, improves the price prediction to consumer's purchase wine
Accuracy.
On the basis of above-described embodiment, the model building module is additionally operable to：
The optimum described many disaggregated models of model parameter are obtained by cross-validation method.
Specifically, because many disaggregated models for building depend on previously given parameter v.Cross validation is adopted in the process
Method choosing optimized parameter, to ensure the classification accuracy of model.The process approach is as follows：By all of consumer's sample set
In each class sample be all divided into 5 parts, each apoplexy due to endogenous wind extracts 4 parts therein as training, remaining for testing.By this
Process is repeated 5 times, and the model parameter chosen in 5 times at least 2 is different, selects the corresponding model of accuracy rate highest
Parameter, so that many disaggregated models are optimum.
The embodiment of the present invention is trained to model so as to obtain optimum many disaggregated models by cross-validation method, is improved
The accuracy of prediction.
On the basis of above-described embodiment, the computing module, specifically for：
The corresponding weight of each factor of influence is calculated according to Lasso algorithms.
Specifically, Lasso algorithms are a kind of Shrinkage estimations.It obtains a more refine by constructing penalty function
Model so that it compresses some coefficients, it is zero to concurrently set some coefficients.Therefore the advantage of subset contraction is remained, is a kind of
Process biased estimation with multi-collinearity data.The basic thought of Lasso be regression coefficient absolute value sum be less than one
Under the constraints of individual constant, minimize residual sum of squares (RSS) such that it is able to produce some regression coefficients exactly equal to 0, obtain
To the model that can be explained.Therefore computing module 702 can calculate the corresponding weight of each factor of influence by Lasso algorithms
The embodiment of the present invention calculates the corresponding weight of each factor of influence by Lasso algorithms so that each factor of influence
Feature have more intuitive and interpretability.
On the basis of above-described embodiment, Fig. 8 buys the valency of wine for a kind of consumer that another embodiment of the present invention is provided
Position prediction system structure diagram, as shown in figure 8, the system includes：701 computing module 702, model of extraction module sets up mould
Block 703, model prediction module 704 and screening module 705, wherein：
Screening module 705 is used to obtain the object effects factor of the weight more than predetermined threshold value, wherein described pre-
If threshold value is more than 0.
Specifically, 701 computing module 702 of extraction module, model building module 703, model prediction module 704 and above-mentioned reality
Apply example consistent, here is omitted.The double supporting vector machine models of two classification v are being set up according to the price of the factor of influence and wine
Before, screening module 705 obtains the object effects factor of the weight more than predetermined threshold value from the object effects factor for getting, its
Middle predetermined threshold value is more than 0, and can be adjusted according to practical situation.
The embodiment of the present invention by obtain weight more than predetermined threshold value the object effects factor, by weight minor impact because
Son removes, only the remaining weight large effect factor, simplifies the double supporting vector machine models of two classification v of foundation.
On the basis of the various embodiments described above, described v pair of supporting vector machine model of two classification is at least three.
Due to needing to set up many disaggregated models according to the double supporting vector machine models of two classification v set up, so two classification v are double
Supporting vector machine model is at least three.
A kind of price prognoses system of consumer's purchase wine that the present invention is provided is used to perform said method, its specific reality
Apply mode consistent with the embodiment of method, here is omitted.
The embodiment of the present invention is by setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of two classification v, improves the price prediction to consumer's purchase wine
Accuracy.
Device embodiment described above is only schematic, wherein the unit as separating component explanation can
To be or may not be physically separate, as the part that unit shows can be or may not be physics list
Unit, you can local to be located at one, or can also be distributed on multiple NEs.Which is selected according to the actual needs can
In some or all of module realizing the purpose of this embodiment scheme.Those of ordinary skill in the art are not paying creativeness
Work in the case of, you can to understand and implement.
Through the above description of the embodiments, those skilled in the art can be understood that each embodiment can
By software plus required general hardware platform mode realizing, naturally it is also possible to by hardware.Based on such understanding, on
State the part that technical scheme substantially contributes prior art in other words to embody in the form of software product, should
Computer software product can be stored in a computer-readable storage medium, such as ROM/RAM, magnetic disc, CD etc., including some fingers
Order is used so that a computer equipment (can be personal computer, server, or network equipment etc.) performs each enforcement
Method described in some parts of example or embodiment.
Finally it should be noted that：Above example only to illustrate technical scheme, rather than a limitation；Although
With reference to the foregoing embodiments the present invention has been described in detail, it will be understood by those within the art that：Which still may be used
To modify to the technical scheme described in foregoing embodiments, or equivalent is carried out to which part technical characteristic；
And these modification or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (10)
1. a kind of consumer buys the price Forecasting Methodology of wine, it is characterised in that include：
The corresponding influence factor of consumer's sample set to getting extracts, and the influence factor includes：Qualitative factor, enterprise
Industry Marketing Factors, motive in purchasing factor and individual characteristicss factor, each influence factor at least include a factor of influence；
The corresponding weight of each factor of influence is calculated, and according to the Weight Acquisition object effects factor；
The double supporting vector machine models of two classification v are set up according to the price of the object effects factor and wine, according to described two classification v
Double supporting vector machine models set up many disaggregated models, and many disaggregated models are instructed using consumer's sample set
Practice；
Forecast sample is treated using many disaggregated models to be predicted.
2. method according to claim 1, it is characterised in that described according to the double supporting vector machine models of described two classification v
Set up many disaggregated models and to many disaggregated models, and many disaggregated models are instructed using consumer's sample set
Practice, also include：
The optimum described many disaggregated models of model parameter are obtained by cross-validation method.
3. method according to claim 1, it is characterised in that the corresponding weight of each factor of influence of the calculating, bag
Include：
The corresponding weight of each factor of influence is calculated according to Lasso algorithms.
4. method according to claim 1, it is characterised in that set up according to the price of the factor of influence and wine described
Before the double supporting vector machine models of two classification v, also include：
The object effects factor of the weight more than predetermined threshold value is obtained, wherein the predetermined threshold value is more than 0.
5. the method according to any one of claim 1-4, it is characterised in that the double supporting vector machine models of the two classification v
At least three.
6. a kind of consumer buys the price prognoses system of wine, it is characterised in that include：
Extraction module, extracts for the corresponding influence factor of consumer's sample set to getting, influence factor's bag
Include：Qualitative factor, enterprise marketing factor, motive in purchasing factor and individual characteristicss factor, each influence factor at least include
One factor of influence；
Computing module, for calculating the corresponding weight of each factor of influence, and according to the Weight Acquisition object effects factor；
Model building module, for setting up the double support vector machine moulds of two classification v according to the price of the object effects factor and wine
Type, sets up many disaggregated models according to the double supporting vector machine models of described two classification v, and using consumer's sample set to described
Many disaggregated models are trained；
Model prediction module, is predicted for treating forecast sample using many disaggregated models.
7. system according to claim 6, it is characterised in that the model building module, is additionally operable to：
The optimum described many disaggregated models of model parameter are obtained by cross-validation method.
8. system according to claim 6, it is characterised in that the computing module, specifically for：
The corresponding weight of each factor of influence is calculated according to Lasso algorithms.
9. system according to claim 6, it is characterised in that the system also includes：
Screening module, for obtaining the object effects factor of the weight more than predetermined threshold value, wherein the predetermined threshold value
More than 0.
10. the system according to any one of claim 6-9, it is characterised in that the double supporting vector machine models of the two classification v
At least three.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611041660.XA CN106600312A (en) | 2016-11-11 | 2016-11-11 | Price prediction method and system for wine buying of consumer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611041660.XA CN106600312A (en) | 2016-11-11 | 2016-11-11 | Price prediction method and system for wine buying of consumer |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106600312A true CN106600312A (en) | 2017-04-26 |
Family
ID=58591931
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611041660.XA Pending CN106600312A (en) | 2016-11-11 | 2016-11-11 | Price prediction method and system for wine buying of consumer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106600312A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985368A (en) * | 2018-07-05 | 2018-12-11 | 四川三联新材料有限公司 | A kind of prediction technique most preferably sucked and system based on cigarette electronic equipment |
| CN111444930A (en) * | 2019-01-17 | 2020-07-24 | 上海游昆信息技术有限公司 | Method and device for determining prediction effect of two-classification model |
-
2016
- 2016-11-11 CN CN201611041660.XA patent/CN106600312A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108985368A (en) * | 2018-07-05 | 2018-12-11 | 四川三联新材料有限公司 | A kind of prediction technique most preferably sucked and system based on cigarette electronic equipment |
| CN111444930A (en) * | 2019-01-17 | 2020-07-24 | 上海游昆信息技术有限公司 | Method and device for determining prediction effect of two-classification model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pineda et al. | An integrated MCDM model for improving airline operational and financial performance | |
| Aydin et al. | Coordination of the closed-loop supply chain for product line design with consideration of remanufactured products | |
| Han et al. | Segmentation of telecom customers based on customer value by decision tree model | |
| Khalili-Damghani et al. | A hybrid fuzzy multiple criteria group decision making approach for sustainable project selection | |
| CN105868847A (en) | Shopping behavior prediction method and device | |
| Li et al. | Vessel traffic flow forecasting by RSVR with chaotic cloud simulated annealing genetic algorithm and KPCA | |
| CN104574220A (en) | Power customer credit assessment method based on least square support vector machine | |
| CN104182474A (en) | Method for recognizing pre-churn users | |
| Anbazhagan et al. | Day-ahead deregulated electricity market price classification using neural network input featured by DCT | |
| CN106779808A (en) | Consumer space's behavior analysis system and method in a kind of commercial circle | |
| CN106971200A (en) | A kind of iconic memory degree Forecasting Methodology learnt based on adaptive-migration | |
| CN106600039A (en) | E-commerce commodity information pushing optimization method | |
| CN110084627A (en) | The method and apparatus for predicting target variable | |
| Huertas-García et al. | Statistical and cognitive optimization of experimental designs in conjoint analysis | |
| CN106022511A (en) | Information predicting method and device | |
| Wee et al. | Bi-level vendor–buyer strategies for a time-varying product price | |
| CN106600312A (en) | Price prediction method and system for wine buying of consumer | |
| Chou et al. | Comparison of AHP and fuzzy AHP methods for human resources in science technology (HRST) performance index selection | |
| Rosalina et al. | A Decision Support System for Determining the Best Customer Using the Simple Multi-Attribute Rating Technique (SMART) | |
| Sun | Sports performance prediction based on chaos theory and machine learning | |
| Shi et al. | Boosting conditional logit model | |
| CN108985595A (en) | The move transaction service evaluation method and device mutually commented based on counterparty | |
| Tsai | Dynamic grey platform for efficient forecasting management | |
| CN108537581A (en) | Based on the GMDH energy consumption Time Series Forecasting Methods selectively combined and device | |
| Harahap et al. | Laptop selection decision support system according to buyer criteria with the simple additive weighting method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170426 |