CN107292670A - A kind of electronic product activation amount Forecasting Methodology and a kind of server cluster - Google Patents
A kind of electronic product activation amount Forecasting Methodology and a kind of server cluster Download PDFInfo
- Publication number
- CN107292670A CN107292670A CN201710524370.9A CN201710524370A CN107292670A CN 107292670 A CN107292670 A CN 107292670A CN 201710524370 A CN201710524370 A CN 201710524370A CN 107292670 A CN107292670 A CN 107292670A
- Authority
- CN
- China
- Prior art keywords
- data
- factor
- activation amount
- electronic product
- variation tendency
- 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.)
- Granted
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
Landscapes
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Data Mining & Analysis (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of electronic product activation amount Forecasting Methodology and a kind of server cluster, and method includes carrying out time series respectively to the first activation amount data and factor I data；Data progress temporal regularity after time series is disassembled and obtains the second activation amount data, the second activation amount data include electronic product activation amount not by the trend data that changed with time in the case of factor I data influence；If variation tendency data have temporal regularity, characterization generation factor Ⅱ data are carried out；Prediction is learnt with factor Ⅱ data input preset model respectively with factor data, is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；The predicted value respectively obtained is compared, to determine target elements data or objective cross in prediction factor data or its various combination.The solution of the present invention is predicted according to different marketing activities to electronic product activation amount data, can obtain real-time, rational marketing program.
Description
Technical field
The present invention relates to data processing, more particularly to a kind of electronic product activation amount Forecasting Methodology and a kind of server set
Group.
Background technology
Whether the marketing strategy is effective, it should which at what time, what channel, to deliver which type of advertisement be most effective
'sThese are the major issues for perplexing sales department for a long time, and an outstanding marketing strategy can not only reduce dispensing
Advertisement, and outstanding effect can be reached, bring important value to enterprise.
At this stage, the assessment of marketing validity is counted according to some simple statistical value such as rates of return on investment (ROI) etc.
Calculate, these statistical informations can not refine to the influence to the specific marketing strategy of product, it is simply strong to some marketing channel
Kang Chengdu general assessment.
The content of the invention
The present invention provides a kind of electronic product activation amount Forecasting Methodology and a kind of server cluster, can quantify different marketing
The influence power of activity, and according to market demand situation, formulate optimal marketing program.
The invention provides a kind of electronic product activation amount Forecasting Methodology, including：
Obtain at least one factor I data of the first activation amount data of the first activation amount data and influence；
Time series is carried out respectively to the first activation amount data and at least one described factor I data；
Temporal regularity is carried out to the data after time series to disassemble, and obtains the second activation amount data, second activation
Amount data are not changed with time including electronic product activation amount by the case of at least one described factor I data influence
Gesture data；
Judge whether the variation tendency data have temporal regularity；
If the variation tendency data have temporal regularity, the variation tendency data are subjected to characterization generation
Factor Ⅱ data；
Different predictions is carried out with factor Ⅱ data input preset model respectively with factor data or its various combination
Study, is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；
The predicted value respectively obtained is compared, with the different prediction factor data or its various combination
Determine target elements data or objective cross.
Preferably, also include,
If it is determined that the variation tendency data do not have temporal regularity, then by the different prediction factor data or
Its various combination inputs preset model and learnt respectively, is predicted simultaneously with the electronic product activation amount to the scheduled time
Respectively obtain predicted value, the predicted value respectively obtained be compared, with the different prediction factor data or its not
With determination target elements data or objective cross in combination.
Preferably, at least one described factor I data, which are selected from, includes Price factor data, marketing activity factor number
According to, one or more of product quality factor data, public opinion factor data, competing product factor data.
Preferably, the public opinion factor data includes affection index factor data, the affection index factor data base
Confirm in positive public opinion evaluation information quantity and negative public opinion evaluation information quantity.
Preferably, carrying out time seriesization at least one described factor I data includes entering Price factor data
Row time series, and before time series is carried out to the Price factor data, the Price factor data are carried out
Discretization.
Preferably, by the variation tendency data progress characterization include, by the variation tendency data according to
Bad duration is configured to one-dimensional characteristic or multidimensional characteristic as the factor Ⅱ data.
Preferably, the prediction factor data is equal with the weight of the factor Ⅱ data.
The invention also discloses a kind of server cluster, including at least one processor, at least one memory, it is described extremely
A few memory can be stored by the instruction of at least one described processor processing, and at least one described processor is configured to hold
Row it is described instruction with：
Obtain at least one factor I data of the first activation amount data of the first activation amount data and influence；
Time series is carried out respectively to the first activation amount data and at least one described factor I data；
Temporal regularity is carried out to the data after time series to disassemble, and obtains the second activation amount data, second activation
Amount data are not changed with time including electronic product activation amount by the case of at least one described factor I data influence
Gesture data；
Judge whether the variation tendency data have temporal regularity；
If the variation tendency data have temporal regularity, the variation tendency data are subjected to characterization generation
Factor Ⅱ data；
Different predictions is carried out with factor Ⅱ data input preset model respectively with factor data or its various combination
Study, is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；
The predicted value respectively obtained is compared, with the different prediction factor data or its various combination
Determine target elements data or objective cross.
Preferably, at least one described processor be configured to further to perform the instruction with：
If it is determined that the variation tendency data do not have temporal regularity, then by the different prediction factor data or
Its various combination inputs preset model and learnt respectively, is predicted simultaneously with the electronic product activation amount to the scheduled time
Respectively obtain predicted value, the predicted value respectively obtained be compared, with the different prediction factor data or its not
With determination target elements data or objective cross in combination.
Preferably, at least one described factor I data, which are selected from, includes Price factor data, marketing activity factor number
According to, one or more of product quality factor data, public opinion factor data, competing product factor data.
Preferably, the public opinion factor data includes affection index factor data, the affection index factor data base
Confirm in positive public opinion evaluation information quantity and negative public opinion evaluation information quantity.
Preferably, carrying out time seriesization at least one described factor I data includes entering Price factor data
Row time series, and before time series is carried out to the Price factor data, the Price factor data are carried out
Discretization.
Preferably, by the variation tendency data progress characterization include, by the variation tendency data according to
Bad duration is configured to one-dimensional characteristic or multidimensional characteristic as the factor Ⅱ data.
Preferably, the prediction factor data is equal with the weight of the factor Ⅱ data.
Compared with prior art, the beneficial effects of the present invention are：It is intelligent due to electronic product so that it can be by
The activation data of electronic product are reported, and the activation data of electronic product can in real time and really react market to electronic product
Changes in demand, and the change for locating different marketing program electronic product demands can be reflected, using reflecting the market demand
Electronic product activation amount data are predicted by electronic product activation amount data according to different marketing activities, can obtain in real time,
Rational marketing program.
Brief description of the drawings
Fig. 1 is the flow chart of the electronic product activation amount Forecasting Methodology of one embodiment of the invention；
Fig. 2 is the flow chart of the electronic product activation amount Forecasting Methodology of another embodiment of the present invention.
Embodiment
The present invention is described in further detail with specific embodiment below in conjunction with the accompanying drawings, but not as the limit to the present invention
It is fixed.
The various schemes and feature of the present invention are described herein with reference to accompanying drawing.
It should be understood that various modifications can be made to the embodiment invented herein.Therefore, description above should not be regarded
To limit, and only as the example of embodiment.Those skilled in the art will expect within the scope and spirit of
Other modifications.
Comprising in the description and constituting the accompanying drawing of a part of specification and show embodiments of the invention, and with it is upper
Substantially description and the detailed description given below to embodiment to the present invention that face is provided are used to explain the present invention together
Principle.
It is of the invention by description with reference to the accompanying drawings to the preferred form of the embodiment that is given as non-limiting examples
These and other characteristic will become apparent.
It is also understood that although with reference to some instantiations, invention has been described, people in the art
Member realize with can determine the present invention many other equivalents, they have feature as claimed in claim and therefore all
In the protection domain limited whereby.
When read in conjunction with the accompanying drawings, in view of described further below, in terms of above and other of the invention, feature and advantage will become
It is more readily apparent.
The specific embodiment of the present invention is described hereinafter with reference to accompanying drawing；It will be appreciated, however, that the embodiment invented is only
The example of the present invention, it can be implemented using various ways.The function and structure known and/or repeated is not described in detail to avoid
Unnecessary or unnecessary details make it that the present invention is smudgy.Therefore, the specific structural and feature invented herein is thin
Section is not intended to restrictions, but as just the basis of claim and representative basis for instruct those skilled in the art with
Substantially any appropriate detailed construction is diversely using the present invention.
This specification can be used phrase " in one embodiment ", " in another embodiment ", " in another embodiment
In " or " in other embodiments ", it may refer to according to one or more of identical or different embodiment of the present invention.
As shown in figure 1, in one embodiment of the present of invention, a kind of electronic product activation amount Forecasting Methodology, including：
S1, obtains at least one factor I data of the first activation amount data of the first activation amount data and influence.
Wherein, the first activation amount data are the data of the user activated electronic product of statistics, and it is user's actual purchase
One reflection of the quantity of electronic product, the quantity of the electronic product of user's actual purchase is more, and the first activation amount data are bigger,
The quantity of user's actual purchase electronic product can be judged according to the first activation amount data, that is, can be according to the first activation
The quantity for the electronic product for measuring data to judge sale.
Price factor data, the marketing activity can be included by influenceing at least one factor I data of the first activation amount data
One or more of factor data, product quality factor data, public opinion factor data, competing product factor data.Price because
Subdata can be the price of electronic product as defined in producer or the price for selling platform, and Price factor data can be with
It is change, for example, producer makes a price reduction for oversold electronic product to electronic product.Marketing activity factor data can be
The human cost or material resources cost that are spent in advertising campaign or represent the fee for showing up of famous person in activity and use or represent place
Lease expenses or represent the spectator attendance in place.Product quality factor data can include repair rate.Public opinion factor data
The discussion temperature or affection index factor data of electronic product can be included.In addition, factor I data can also include it is competing
Product activity factor data, competing product activity factor data can be done advertising campaign as the producer of rival or represent activity
Data.
First activation amount data and at least one factor I data are carried out time series by S2 respectively.
It can be that the first activation amount data are carried out sequentially in time to carry out time seriesization to the first activation amount data
Arrangement, for example, arrangement data of the activation amount of electronic product according to the date.
It can be that factor I data are arranged sequentially in time to carry out time seriesization to factor I data.
For example, it can be arrangement data of the price as defined in producer according to the date to carry out time seriesization to Price factor data, if
Producer embodies in certain section of time reduction price, Price factor data that can be after time series.
In addition, can be averagely to generation by fee for showing up for the time seriesization for representing the fee for showing up of famous person in activity
Speech activity during every day, according still further to this section during in every day correspondence equalization after fee for showing up use.Lease expenses or
The time seriesization of spectator attendance is similar to the time series of fee for showing up, will not be repeated here.For competing product activity factor
It is also similar therewith that data carry out time seriesization.
Time seriesization for repair rate can be arrangement data of the repair rate according to the date.And for discussing temperature
Time seriesization can be captured people in online discussion using crawler technology, and calculate begging on electronic product
The quantity of opinion, finally correspond to the date arranged.Can be based on for the affection index factor data in public opinion factor data
Positive public opinion evaluation information quantity and negative public opinion evaluation information quantity confirm.Can be specifically, by front and negatively
Public opinion evaluation information counted, and according to (nJust-nIt is negative)*10/(nJust+nIt is negative) calculate and obtain affection index, and to affection index
Arranged according to the time.Wherein, nJustFor the quantity of the positive public opinion evaluation information on the same day of statistics, nIt is negativeFor same day of statistics
The quantity of negative public opinion evaluation information.
Data after time series are carried out temporal regularity and disassembled, obtain the second activation amount data, the second activation amount by S3
Data include electronic product activation amount not by the trend data that changed with time in the case of at least one factor I data influence.
Wherein, it can utilize Time series analysis method pair the data after time series to be carried out temporal regularity to disassemble
Data after time series carry out temporal regularity and disassembled, specifically, can utilize the STL in Time series analysis method
(Seasonal and Trend decomposition using Loess ') decomposition method, STL decomposition methods are with robust office
Portion's weighted regression as smoothing method Time Series method, so as to obtain the second activation amount data, wherein second
Activation amount data are not changed with time including electronic product activation amount by the case of at least one factor I data influence
Gesture data, that is to say, the second activation amount data are after the influence of factor I data is peeled off from the first activation amount data
The data of the activation amount of the electronic product changed over time, the trend data changed over time becomes for a long time including what is changed over time
Gesture data and the short-term trend data changed over time.Wherein, the long-term trend data changed over time can be electronic product
The long-term trend data that activation amount is changed over time, for example, the long-term trend data changed over time in 1 year.With the time
The short-term trend data of change can be the short-term trend data that electronic product activation amount changes over time, for example, 7 days it
Interior change.
S4, judges whether the variation tendency data have temporal regularity.Judge whether variation tendency data have the time
Rule can be specifically whether sometimes to judge the long-term trend data changed over time or the short-term trend data changed over time
Between rule, i.e. judge whether the long-term trend data that change over time gradually reduce or gradually rise, or change over time
It is periodically or seasonal whether short-term trend data have.
Variation tendency data, if variation tendency data have temporal regularity, are carried out characterization generation second by S5
Factor data.
If it is determined that, the long-term trend data changed over time gradually rise, for example in 1 year over time
And gradually rise, the short-term trend data changed over time are to gradually change in the cycle for example with 7 days, then, change is become
Gesture data carry out characterization generation factor Ⅱ data.
Wherein, variation tendency data characterize to can include will the long-term trend data that change over time and at any time
Between the dependence duration of short-term trend data that changes be configured to one-dimensional characteristic or multidimensional characteristic as factor Ⅱ data.For example,
If the short-term trend data changed over time with 7 days be the cycle gradually change, then will with 7 days be periodical input for example
Lag (n) functions are configured to 7 degree of freedom feature.In another example, if the short-term trend data changed over time are for the cycle with 1 day
Gradually change, then will be that periodical input such as Lag (n) functions are configured to one-dimensional characteristic with 1 day.
S6, different prediction factor data or its various combination are entered with factor Ⅱ data input preset model respectively
Row study, is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value.
Factor Ⅱ data are learnt from different predictions with factor data or its various combination input preset model,
It is predicted with the electronic product activation amount to the same scheduled time.Preset model can be integrated study model, be, for example,
GBDT or XGBOOST algorithms.By factor I data and different prediction factor datas or the default mould of its various combination input
During type, each prediction factor data is equal with the weight of factor Ⅱ data or the predictive factor data that combine in
Each predictive factor data is equal with the weight of factor Ⅱ data.
Activation amount to electronic product is predicted, for example, can be that will change over time the long-term trend spy being gradually increasing
Levying and be factor Ⅱ data after being characterized with 7 days short-term trend data for the cycle, and by the scheduled time and
Different predictions is learnt with factor data or its various combination input preset model, different predictions factor number now
According to or its various combination can from such as under on-line selling, line sell, deliver billboard different channels, for example price reduction pin
Sell, bundle sale, buy the different marketing modes that electronic product gives peripheral hardware etc. as an addition, the difference such as representing activity, show activity
Selected in marketing dynamics, so as to obtain in factor Ⅱ data and different prediction factor datas or its various combination work
Under, the electronic product activation amount of the same scheduled time, so as to obtain different prediction factor data or its different groups
Close the influence to electronic product activation amount data.Wherein, the scheduled time can be the time or future of following some day
A certain of time.Different prediction factor data for different marketing modes, different channel, different marketing dynamics etc. or
Its various combination can carry out characterization and input preset model again being learnt.
S7, is compared to the predicted value respectively obtained, with different prediction factor datas or its various combination
Determine target elements data or objective cross.
If for example, respectively to electronic product carry out spend 500,000 or 300,000 represent activity as different predictions because
Subdata, the data of obtained electronic product activation amount are more or less the same, then 300,000 activity of representing can be selected to be used as marketing side
Case.
Using the electronic product activation amount data for reflecting the market demand, according to different marketing activities to electronic product activation amount
Data are predicted, and can obtain real-time, rational marketing program.
As shown in Fig. 2 in another embodiment disclosed by the invention, a kind of electronic product activation amount Forecasting Methodology, including：
S1, obtains at least one factor I data of the first activation amount data of the first activation amount data and influence；
First activation amount data and at least one factor I data are carried out time series by S2 respectively；
Data after time series are carried out temporal regularity and disassembled, obtain the second activation amount data, the second activation amount by S3
Data include electronic product activation amount not by the trend data that changed with time in the case of at least one factor I data influence；
S4, judges whether variation tendency data have temporal regularity；
S8, if it is decided that variation tendency data do not have temporal regularity, then by different prediction factor datas or its not
With combination, input preset model is learnt respectively, is predicted and is respectively obtained with the electronic product activation amount to the scheduled time
Predicted value, is compared to the predicted value respectively obtained, to be determined in different prediction factor datas or its various combination
Target elements data or objective cross.
The present embodiment and the difference of upper one embodiment are essentially consisted in S8, S8, if it is decided that variation tendency data do not have
There is temporal regularity, then different predictions is inputted into preset model respectively with factor data or its various combination is learnt, with right
The electronic product activation amount of the same scheduled time is predicted and respectively obtains predicted value, it is thus possible to be advised without the time
Different predictions is predicted with factor data or its various combination in the case of the variation tendency data of rule, to respectively obtaining
Predicted value be compared, to determine target elements data or target in different prediction factor datas or its various combination
Combination.Preset model can be integrated study model, e.g. GBDT or XGBOOST algorithms.Wherein, the scheduled time can be not
Carry out the time or following a certain time of some day.
In the above two embodiments, carrying out time seriesization at least one factor I data is included to Price factor
Data carry out time series, and to Price factor data carry out time series before, Price factor data are carried out from
Dispersion, for example, it is interval discrete to price data can to turn to different price lattice.
The invention also discloses a kind of server cluster, including at least one processor, at least one memory, at least one
Individual memory can store the instruction handled by least one processor, at least one processor be configured to execute instruction with：
Obtain at least one factor I data of the first activation amount data of the first activation amount data and influence；
Time series is carried out respectively to the first activation amount data and at least one factor I data；
Temporal regularity is carried out to the data after time series to disassemble, and obtains the second activation amount data, the second activation amount number
According to including electronic product activation amount not by the trend data that changed with time in the case of at least one factor I data influence；
Judge whether variation tendency data have temporal regularity；
If variation tendency data have temporal regularity, variation tendency data are subjected to characterization generation factor Ⅱ
Data；
Different predictions is carried out with factor Ⅱ data input preset model respectively with factor data or its various combination
Study, is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；
The predicted value respectively obtained is compared, to be determined in different prediction factor datas or its various combination
Target elements data or objective cross.
Further, at least one processor be configured to further execute instruction with：
If it is determined that variation tendency data do not have temporal regularity, then by different prediction factor datas or its different groups
Close input preset model respectively to be learnt, be predicted with the electronic product activation amount to the scheduled time and respectively obtain prediction
Value, is compared to the predicted value respectively obtained, to determine target with factor data or its various combination in different predictions
Factor data or objective cross.
Further, at least one factor I data, which is selected from, includes Price factor data, marketing activity factor data, production
One or more of quality factor data, public opinion factor data, competing product factor data.
Further, public opinion factor data includes affection index factor data, and affection index factor data is based on positive carriage
Confirm by evaluation information quantity and negative public opinion evaluation information quantity.
Further, when carrying out time seriesization including being carried out to Price factor data at least one factor I data
Between serialize, and to Price factor data carry out time series before, to Price factor data carry out discretization.
Further, variation tendency data progress characterization is included, by the dependence duration structure of variation tendency data
One-dimensional characteristic or multidimensional characteristic are caused as factor Ⅱ data.
Further, prediction factor data is equal with the weight of factor Ⅱ data.
Using the electronic product activation amount data for reflecting the market demand, according to different marketing activities to electronic product activation amount
Data are predicted, and can obtain real-time, rational marketing program.
Above example is only the exemplary embodiment of the present invention, is not used in the limitation present invention, protection scope of the present invention
It is defined by the claims.Those skilled in the art can make respectively in the essence and protection domain of the present invention to the present invention
Modification or equivalent substitution are planted, this modification or equivalent substitution also should be regarded as being within the scope of the present invention.
Claims (14)
1. a kind of electronic product activation amount Forecasting Methodology, it is characterised in that including：
Obtain at least one factor I data of the first activation amount data of the first activation amount data and influence；
Time series is carried out respectively to the first activation amount data and at least one described factor I data；
Temporal regularity is carried out to the data after time series to disassemble, and obtains the second activation amount data, the second activation amount number
According to including electronic product activation amount not by the trend number that changed with time in the case of at least one described factor I data influence
According to；
Judge whether the variation tendency data have temporal regularity；
If the variation tendency data have temporal regularity, the variation tendency data are subjected to characterization generation second
Factor data；
Different predictions is learnt with factor Ⅱ data input preset model respectively with factor data or its various combination,
It is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；
The predicted value respectively obtained is compared, to be determined in the different prediction factor data or its various combination
Target elements data or objective cross.
2. according to the method described in claim 1, it is characterised in that also include,
If it is determined that the variation tendency data do not have temporal regularity, then by the different prediction factor data or its not
With combination, input preset model is learnt respectively, is predicted and is distinguished with the electronic product activation amount to the scheduled time
Predicted value is obtained, the predicted value respectively obtained is compared, with the different prediction factor data or its different groups
Target elements data or objective cross are determined in conjunction.
3. method according to claim 1 or 2, it is characterised in that at least one described factor I data, which are selected from, to be included
Price factor data, marketing activity factor data, product quality factor data, public opinion factor data, competing product factor data
One or more of.
4. method according to claim 3, it is characterised in that the public opinion factor data includes affection index factor number
According to the affection index factor data is based on positive public opinion evaluation information quantity and negative public opinion evaluation information quantity confirms.
5. method according to claim 1 or 2, it is characterised in that when being carried out at least one described factor I data
Between serializing include to Price factor data carry out time series, and to the Price factor data carry out time series
Before change, discretization is carried out to the Price factor data.
6. method according to claim 1 or 2, it is characterised in that the variation tendency data are subjected to characterization
Including the dependence duration of the variation tendency data to be configured to one-dimensional characteristic or multidimensional characteristic as the factor Ⅱ number
According to.
7. according to the method described in claim 1, it is characterised in that the prediction factor data and the factor Ⅱ data
Weight it is equal.
8. a kind of server cluster, it is characterised in that including at least one processor, at least one memory, described at least one
Individual memory can be stored by the instruction of at least one described processor processing, and at least one described processor is configured to perform institute
State instruction with：
Obtain at least one factor I data of the first activation amount data of the first activation amount data and influence；
Time series is carried out respectively to the first activation amount data and at least one described factor I data；
Temporal regularity is carried out to the data after time series to disassemble, and obtains the second activation amount data, the second activation amount number
According to including electronic product activation amount not by the trend number that changed with time in the case of at least one described factor I data influence
According to；
Judge whether the variation tendency data have temporal regularity；
If the variation tendency data have temporal regularity, the variation tendency data are subjected to characterization generation second
Factor data；
Different predictions is learnt with factor Ⅱ data input preset model respectively with factor data or its various combination,
It is predicted with the electronic product activation amount to the scheduled time and respectively obtains predicted value；
The predicted value respectively obtained is compared, to be determined in the different prediction factor data or its various combination
Target elements data or objective cross.
9. server cluster according to claim 8, it is characterised in that at least one described processor is configured to further
Perform it is described instruction with：
If it is determined that the variation tendency data do not have temporal regularity, then by the different prediction factor data or its not
With combination, input preset model is learnt respectively, is predicted and is distinguished with the electronic product activation amount to the scheduled time
Predicted value is obtained, the predicted value respectively obtained is compared, with the different prediction factor data or its different groups
Target elements data or objective cross are determined in conjunction.
10. server cluster according to claim 8 or claim 9, it is characterised in that at least one factor I data choosing
From including Price factor data, marketing activity factor data, product quality factor data, public opinion factor data, competing product because
One or more of subdata.
11. server cluster according to claim 10, it is characterised in that the public opinion factor data includes affection index
Factor data, the affection index factor data is based on positive public opinion evaluation information quantity and negative public opinion evaluation information quantity
Confirm.
12. server cluster according to claim 8 or claim 9, it is characterised in that at least one described factor I data
Carrying out time seriesization includes carrying out time series to Price factor data, and when being carried out to the Price factor data
Between serialize before, to the Price factor data carry out discretization.
13. server cluster according to claim 8 or claim 9, it is characterised in that the variation tendency data are subjected to feature
Change processing include, the dependence duration of the variation tendency data is configured to one-dimensional characteristic or multidimensional characteristic as described second because
Subdata.
14. server cluster according to claim 8, it is characterised in that the prediction factor data and described second
The weight of factor data is equal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710524370.9A CN107292670B (en) | 2017-06-30 | 2017-06-30 | Electronic product activation amount prediction method and server cluster |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710524370.9A CN107292670B (en) | 2017-06-30 | 2017-06-30 | Electronic product activation amount prediction method and server cluster |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107292670A true CN107292670A (en) | 2017-10-24 |
| CN107292670B CN107292670B (en) | 2020-11-20 |
Family
ID=60098387
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710524370.9A Active CN107292670B (en) | 2017-06-30 | 2017-06-30 | Electronic product activation amount prediction method and server cluster |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107292670B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050256759A1 (en) * | 2004-01-12 | 2005-11-17 | Manugistics, Inc. | Sales history decomposition |
| CN103617548A (en) * | 2013-12-06 | 2014-03-05 | 李敬泉 | Medium and long term demand forecasting method for tendency and periodicity commodities |
| CN106408341A (en) * | 2016-09-21 | 2017-02-15 | 北京小米移动软件有限公司 | Goods sales volume prediction method and device, and electronic equipment |
| CN106648709A (en) * | 2017-01-10 | 2017-05-10 | 深圳铂睿智恒科技有限公司 | Method and system for analyzing promotion revenue data of preinstalled applications of smart terminals |
| CN106845683A (en) * | 2016-12-26 | 2017-06-13 | 南通同心烟智道文化传播有限公司 | Big data tobacco Method for Sales Forecast method based on the network platform |
-
2017
- 2017-06-30 CN CN201710524370.9A patent/CN107292670B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050256759A1 (en) * | 2004-01-12 | 2005-11-17 | Manugistics, Inc. | Sales history decomposition |
| CN103617548A (en) * | 2013-12-06 | 2014-03-05 | 李敬泉 | Medium and long term demand forecasting method for tendency and periodicity commodities |
| CN106408341A (en) * | 2016-09-21 | 2017-02-15 | 北京小米移动软件有限公司 | Goods sales volume prediction method and device, and electronic equipment |
| CN106845683A (en) * | 2016-12-26 | 2017-06-13 | 南通同心烟智道文化传播有限公司 | Big data tobacco Method for Sales Forecast method based on the network platform |
| CN106648709A (en) * | 2017-01-10 | 2017-05-10 | 深圳铂睿智恒科技有限公司 | Method and system for analyzing promotion revenue data of preinstalled applications of smart terminals |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107292670B (en) | 2020-11-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Bajari et al. | Economic insights from internet auctions | |
| Dasu et al. | Dynamic pricing when consumers are strategic: Analysis of posted and contingent pricing schemes | |
| Zhao et al. | Dynamic pricing in the presence of consumer inertia | |
| Nosoohi et al. | Outsource planning through option contracts with demand and cost uncertainty | |
| KR20180041405A (en) | Apparatus for providing sales forecasting information based on internet | |
| Jiang et al. | Optimizing shipping-fee schedules to maximize e-tailer profits | |
| Priyan et al. | Trade credit financing in the vendor–buyer inventory system with ordering cost reduction, transportation cost and backorder price discount when the received quantity is uncertain | |
| Plott et al. | Call market experiments: Efficiency and price discovery through multiple calls and emergent Newton adjustments | |
| Ruffle | When do large buyers pay less? Experimental evidence | |
| Dass et al. | Maximizing bidder surplus in simultaneous online art auctions via dynamic forecasting | |
| CN107248094A (en) | A kind of electronic product activation amount Forecasting Methodology and a kind of server cluster | |
| Moulet et al. | The influence of seller learning and time constraints on sequential bargaining in an artificial perishable goods market | |
| CN107292670A (en) | A kind of electronic product activation amount Forecasting Methodology and a kind of server cluster | |
| Guerci et al. | Learning to bid in sequential Dutch auctions | |
| Lu et al. | Information transparency in B2B auction markets: The role of winner identity disclosure | |
| Kim | Maximizing sellers’ welfare in online auction by simulating bidders’ proxy bidding agents | |
| Ruffle | Buyer countervailing power: a survey of the theory and experimental evidence | |
| CN107330732A (en) | A kind of electronic product pricing method and a kind of server cluster | |
| Shah et al. | Optimal inventory policies for non-instantaneous deteriorating items with advance sales and advertisement efforts | |
| KR102324553B1 (en) | Smart system for mediating in sale based on predictive consumer | |
| CN111667314A (en) | Commodity resource allocation method and device and electronic equipment | |
| Dass et al. | Dynamic price forecasting in simultaneous online art auctions | |
| Lu et al. | An empirical model for multi-unit sequential Dutch auctions: the Dutch flower auctions case | |
| Čivić | Customer relationship management in business operations of trading companies in Bosnia and Herzegovina | |
| Jalaly et al. | Learning and trust in auction markets |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |