CN106845682A - The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released - Google Patents
The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released Download PDFInfo
- Publication number
- CN106845682A CN106845682A CN201611191944.7A CN201611191944A CN106845682A CN 106845682 A CN106845682 A CN 106845682A CN 201611191944 A CN201611191944 A CN 201611191944A CN 106845682 A CN106845682 A CN 106845682A
- Authority
- CN
- China
- Prior art keywords
- winning
- bid
- rate
- bidding
- knock
- 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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0611—Request for offers or quotes
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Marketing (AREA)
- Theoretical Computer Science (AREA)
- Human Resources & Organizations (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released, methods described comprise the following steps：Obtain bid results of the information bit to be released on full flow, the full flow includes representing request for the multiple of the information bit to be released, each to represent request and include corresponding bid, the multiple request covering multiple that represents is bid sections；According to the knock-down price of the information bit to be released when being won every time in the bid results, determine that the knock-down price and the knock-down price of the rate of winning of bidding are bidded rate relation curve of winning；The rate relation curve prediction of winning of being bidded based on the knock-down price is bidded the rate of winning.The technical scheme provided by the present invention, can bid win rate of the more intuitive and accurate prediction information bit to be released on full flow, and then be conducive to party in request's platform bid is carried out much sooner, accurately adjustment.
Description
Technical field
The present invention relates to internet data issue field, more particularly to a kind of rate of winning of bidding of information bit to be released
Forecasting Methodology and device.
Background technology
Existing information issue market is entered using real time bid (real time bidding, abbreviation RTB) pattern mostly
The transaction of row information to be released, so as to be estimated and go out for the displaying behavior of each user using precision marketing technology
Valency.In this information trading pattern to be released, whole Information Ecochain to be released includes information master to be released, party in request's platform
(demand side platform, abbreviation DSP), four main bodys of information trading platform to be released and the Internet media, wherein,
Be put into oneself information issue demand on party in request's platform by information publisher, and the Internet media is by the information flow to be released of oneself
Amount resource is put into information trading platform to be released, by party in request's platform by complete with the interface differential technique of information trading platform to be released
Into the purchase of bidding of information bit to be released.
In party in request's platform at this stage, the more commonly used method is by clicking on arrival rate (click through
Rate, abbreviation CTR) carry out the bidding strategy of Nutrition guide needs side's platform.The click arrival rate can be understood as information to be released
Result of the clicking rate divided by the amount of representing of information to be released.Party in request's platform is based on clicking on arrival rate, by what is raised the price or reduce the price
Method is bid to improve the rate of winning of bidding to adjust.
But, based on such scheme, party in request's platform intuitive and accurate cannot know that bidding for information bit to be released is won
The actual conditions of rate, and then it is unfavorable for timely, accurate adjustment of party in request's platform to bidding.
The content of the invention
Present invention solves the technical problem that being that prior art cannot be so that more intuitively method carrys out Accurate Prediction letter to be released
Bid win rate of the breath position on full flow, is unfavorable for that party in request's platform carries out timely, accurate adjustment to bid.
In order to solve the above technical problems, the embodiment of the present invention provides a kind of prediction of the rate of winning of bidding of information bit to be released
Method, comprises the following steps：Bid results of the information bit to be released on full flow are obtained, the full flow includes being directed to
The multiple of the information bit to be released represents request, each to represent request and include corresponding bid, the multiple to represent request and cover
The multiple bid sections of lid；According to the knock-down price of the information bit to be released when being won every time in the bid results, it is determined that described
Knock-down price-rate the relation curve of winning of bidding of knock-down price and the rate of winning of bidding；Based on the knock-down price-rate relation song of winning of bidding
Line predicts the rate of winning of bidding.
Optionally, bid results of the information bit to be released on full flow are obtained, is comprised the following steps：For described
Information bit to be released, the bid for representing request and corresponding bid results at least one bid section known to statistics；Root
According to described at least one bid section in the bid for representing request and corresponding bid results, fitting obtain it is described bid with it is competing
Bid-rate relation curve of winning of bidding that valency is won between rate；Based on the bid-bid and win described in the determination of rate relation curve
The rate of winning of bidding is 100% bid section；With bid section that the rate of winning of bidding is 100% as the upper limit, by default
The random bid detection of number of times, to obtain the distribution of bid win rate of the information bit to be released on full flow.
Optionally, the bid-bid rate relation curve of winning is based on following steps and is fitted to obtain：According to Gaussian function with
And error function fitting obtains the bid-rate relation curve of winning of bidding.
Optionally, according in the bid results every time win when the information bit to be released knock-down price determine it is described into
Friendship valency and the knock-down price-rate relation curve of winning of bidding of the rate of winning of bidding, comprise the following steps：According to the random bid detection
Result, obtain the knock-down price and corresponding number of times of winning of the information bit to be released when winning every time；For each conclusion of the business
Valency, the rate of winning of bidding for obtaining the knock-down price is calculated based on the corresponding number of times of winning of the knock-down price；According to the knock-down price
And corresponding rate of winning of bidding determines to obtain the knock-down price-rate relation curve of winning of bidding.
Optionally, the knock-down price-bid rate relation curve of winning is determined based on following steps：According to Gaussian function and
Error function fitting obtains the knock-down price-rate relation curve of winning of bidding.
Optionally, it is described that the knock-down price-rate relation of winning of bidding is obtained based on Gaussian function and error function fitting
Curve, comprises the following steps：Calculated based on equation below and obtain the knock-down price-rate relation curve of winning of bidding：F (x)=ω1
(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+erf (x)), wherein, f (x) is the knock-down price-bid and win
Rate relation curve, ωiIt is i-th weight distribution of Gaussian function, x is the rate of winning of bidding, and erf (x) is error function, and n is pre-
If Gauss curve fitting quantity, 1≤i≤n.
Optionally, the error function is based on equation below and calculates acquisition：
Wherein, erf (x) is the error function, and x is the rate of winning of bidding, and t is Gaussian parameter；The Gaussian parameter is based on formula t=
(x-m)/(2 × δ) calculate and obtain, wherein, x is the rate of winning of bidding, and m is the average of the normal distribution of Gaussian function, and δ is height
The covariance of the normal distribution of this function.
The embodiment of the present invention also provides a kind of prediction meanss of the rate of winning of bidding of information bit to be released, including：Obtain mould
Block, for obtaining bid results of the information bit to be released on full flow, the full flow is included for described to be released
The multiple of information bit represents request, each to represent request and include corresponding bid, the multiple to represent that request covering is multiple to bid
Section；Determining module, for the knock-down price according to the information bit to be released when being won every time in the bid results, determines institute
State the knock-down price-rate relation curve of winning of bidding of knock-down price and the rate of winning of bidding；Prediction module, for based on the knock-down price-
The rate relation curve prediction of winning of bidding is bidded the rate of winning.
Optionally, the acquisition module includes：Statistic submodule, for the information bit to be released, known to statistics extremely
The bid for representing request and corresponding bid results in few bid section；First fitting submodule, for according to described
The bid for representing request and corresponding bid results at least one bid section, fitting obtain the bid and win with bidding
Bid-rate relation curve of winning of bidding between rate；Determination sub-module, for based on the bid-the rate relation of winning of bidding is bent
Line is bidded the bid section that the rate of winning is 100% described in determining；First treatment submodule, for being with the rate of winning of bidding
100% bid section is the upper limit, is detected by the random bid of preset times, is being flowed entirely with obtaining the information bit to be released
The distribution of the rate of winning of bidding in amount.
Optionally, the first fitting submodule includes：Fitting unit, for being intended according to Gaussian function and error function
Close and obtain the bid-rate relation curve of winning of bidding.
Optionally, the determining module includes：Acquisition submodule, for the result according to the random bid detection, obtains
Take the knock-down price and corresponding number of times of winning of the information bit to be released when winning every time；Calculating sub module, for it is each into
Friendship valency, the rate of winning of bidding for obtaining the knock-down price is calculated based on the corresponding number of times of winning of the knock-down price；Second processing submodule
Root tuber determines to obtain the knock-down price-rate relation curve of winning of bidding according to the knock-down price and corresponding rate of winning of bidding.
Optionally, the determining module includes：Second fitting submodule, for being intended according to Gaussian function and error function
Close and obtain the knock-down price-rate relation curve of winning of bidding.
Optionally, the second fitting submodule includes：First computing unit, institute is obtained for being calculated based on equation below
State knock-down price-rate relation curve of winning of bidding：F (x)=ω1(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+
Erf (x)), wherein, f (x) is the knock-down price-rate relation curve, ω of winning of biddingiIt is i-th weight of Gaussian function point
Cloth, x is the rate of winning of bidding, and erf (x) is error function, and n is default Gauss curve fitting quantity, 1≤i≤n.
Optionally, the second fitting submodule also includes：Second computing unit, is obtained for being calculated based on equation below
The error function：Wherein, erf (x) is the error function, and x is victory of bidding
Extracting rate, t is Gaussian parameter；3rd computing unit, the Gauss ginseng is obtained for being calculated based on formula t=(x-m)/(2 × δ)
Number, wherein, x is the rate of winning of bidding, and m is the average of the normal distribution of Gaussian function, and δ is the normal distribution of Gaussian function
Covariance.
Compared with prior art, the technical scheme of the embodiment of the present invention has the advantages that：
Bid results of the information bit to be released on full flow are obtained, during according to being won every time in the bid results
The knock-down price of the information bit to be released determines knock-down price-rate relation curve of winning of bidding, so that based on the knock-down price-competing
Valency rate relation curve of winning predicts the rate of winning of bidding of the information bit to be released on full flow.Than existing based on bid
It is fitted to predict the technical scheme of the rate of winning of bidding, the technical side of the embodiment of the present invention with the corresponding relation of the rate of winning of bidding
Case can be with the knock-down price as starting point, and the more intuitive and accurate rate of winning of bidding to the information bit to be released is carried out
Prediction, is conducive to Nutrition guide needs side's platform to adjust bid in time, improves the number of times of winning of the information bit to be released.
Further, the bid for representing request at least one bid section known to statistics and corresponding bid results,
So as to be fitted rate relation curve of winning of being bid-bidded, and based on the bid-the rate relation curve of winning of bidding determines to bid
The rate of winning is 100% bid section, so as to bid section that the rate of winning of bidding is 100% as the upper limit makees preset times
Random bid detection, to obtain the distribution of bid win rate of the information bit to be released on full flow.Than existing skill
Art can only predict the technical scheme of the rate of winning of bidding based on history bid and corresponding bid results, the embodiment of the present invention
Technical scheme preferably carries out preset times to understand the influence that different bid sections may be caused to the rate of winning of bidding
Bid detection, especially bid-bid to win the bid area that the rate of winning is 100% that bidded determined by rate relation curve with early stage
Section on the basis of the upper limit, to carry out high price detection, to obtain bid results of the information bit to be released on full flow so that
More comprehensive fitting sample is provided to fit the knock-down price-bid rate relation curve of winning.
Brief description of the drawings
Fig. 1 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the first embodiment of the present invention
Figure；
Fig. 2 is the knock-down price-rate relation curve of winning of bidding of the information bit to be released obtained using the embodiment of the present invention；
Fig. 3 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the second embodiment of the present invention
Figure；
Fig. 4 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the third embodiment of the present invention
Figure；
Fig. 5 is the structure of the prediction meanss of the rate of winning of bidding of the information bit a kind of to be released of the fourth embodiment of the present invention
Schematic diagram.
Specific embodiment
It will be appreciated by those skilled in the art that the general divalence auction using closing in practical operation due to real time bid pattern
Method carries out the price auction of information bit to be released, and party in request's platform cannot generally grasp the information bit to be released in full flow
On knock-down price, thus, prior art predict information bit to be released bid win rate when, typically can only be by bid-competing
Valency wins rate relation curve to predict the rate of winning of bidding of the information bit to be released.But, due to divalence auction system be with
Used as the knock-down price of the information bit to be released, this results in party in request's platform and is won with bidding and bidding for second bid high
The bid that rate is fitted for major consideration-bid win rate relation curve can not it is very accurate, intuitively reflect and treat
Influence of the actual knock-down price in position that releases news to the rate of winning of bidding.
On the other hand, the curve-fitting method that existing party in request's platform is used is mostly that least square method etc. is more conventional
Means, but this approximating method cannot ensure curve that fitting obtains for increasing trend, and bid and the rate of winning of bidding
Actual corresponding relation is not corresponded, and is unfavorable for the subsequent operation of party in request's platform.
In order to solve this technical problem, technical scheme of the present invention obtains the information bit to be released on full flow
Bid results, according in the bid results every time win when the information bit to be released knock-down price come determine knock-down price-
Bid rate relation curve of winning, so that based on the knock-down price-the rate relation curve of winning of bidding predicts the information bit to be released
The rate of winning of bidding on full flow such that it is able to the knock-down price as starting point, it is more intuitive and accurate to described pending
The rate of winning of bidding of cloth information bit is predicted, and is conducive to Nutrition guide needs side's platform to adjust bid in time, improves to described pending
The number of times of winning of cloth information bit.
It is understandable to enable above-mentioned purpose of the invention, feature and beneficial effect to become apparent, below in conjunction with the accompanying drawings to this
The specific embodiment of invention is described in detail.
Fig. 1 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the first embodiment of the present invention
Figure.Wherein, the information bit to be released can be used for showing information to be released that the information to be released can be that various push is believed
Breath (such as news), advertisement (such as Internet advertising) etc.；The rate of winning of bidding (win rate) can be described to be released
The probability of winning that information is bidded in the information bit to be released.
It will be appreciated by those skilled in the art that the rate of winning of bidding is close with the bid (biding price) of party in request's platform
Correlation, then it is described bid the rate of winning can by formula wr=a/b, wherein, a for a period of time in it is specific bid go participate in institute
State the number of times of winning after the bidding of information bit to be released, b is to go with identical or different bid to participate in described in described a period of time
The total degree bidded of information bit to be released.Further, for the real time bid pattern using closed divalence auction technique
Information bit to be released, for this situation for winning of bid of party in request's platform, during practical application, the information to be released
The supplier of position may be come and the party in request as knock-down price (win price) using epicycle middle highest price of bidding
Platform settle accounts, in the application scenarios by taking Internet advertising as an example, in typically being bidded using epicycle the second price high as into
Friendship valency (win price) come with party in request's platform settle accounts, then a can also for a period of time in participate in it is described pending
The number of times of winning of specific knock-down price after the bidding of cloth information bit.
Specifically, in the present embodiment, step S101 is first carried out, the acquisition information bit to be released is on full flow
Bid results, the full flow includes representing request for the multiple of the information bit to be released, it is each represent ask to include it is right
The bid answered, it is the multiple to represent the multiple bid sections of request covering.More specifically, the multiple bid section covering is described
The acceptable bid ranges of information bit to be released.Preferably, it is the multiple represent request can with it is the multiple bid section one
One correspondence, for example, representing request for each, its bid for including is the suitable bid from corresponding bid section selection.
In a change case, it is the multiple represent request can not also with it is the multiple bid section correspond, example
Such as, multiple bids that represent request and can include can fall into same bid section, to improve to the information bit to be released
Bid the rate of winning.
Subsequently into step S102 perform, according in the bid results every time win when the information bit to be released into
Friendship valency, determines the knock-down price-rate relation curve of winning of bidding of the knock-down price and the rate of winning of bidding.Specifically, the knot of bidding
Fruit includes bidding successfully still bid unsuccessfully.More specifically, the bid results are bidded knock-down price when successfully including this.
Further, the bid results also include party in request's platform on the full flow of the information bit to be released with same
Bid sends multiple number of times of winning of bidding represented after request, and the corresponding knock-down price of bidding when bidding successfully.It is preferred that
Ground, the knock-down price can be based on actual delivery when party in request's platform is settled accounts with the supplier of the information bit to be released
Price determines.In a preference, corresponding the bidding of identical knock-down price is won in counting the bid results on the full flow
Number of times, counts respective the bidding of each knock-down price and wins number of times in proportion in number of times of winning of always bidding, and the ratio is institute
State the corresponding rate of winning of bidding of knock-down price, and then with the corresponding relation of multiple knock-down prices and the rate of winning of bidding as sample point, fitting
Obtain the knock-down price-rate relation curve of winning of bidding.
Step S103 is finally performed, based on the knock-down price-the rate relation curve prediction of winning of bidding bids the rate of winning.Ability
Field technique personnel understand that the multiple knock-down price obtained in above-mentioned steps S102 is preferred with the corresponding relation of the rate of winning of bidding
For representing multiple sample points that the knock-down price-bid is won on rate relation curve, in addition to these sample points into
Friendship valency, because party in request's platform is not won on corresponding bidding, causes party in request's platform to have no way of knowing correspondence
The rate of winning of bidding so that be fitted on the basis of the multiple sample point obtaining the knock-down price-rate pass of winning of bidding
Be curve so that party in request's platform can predict the rate of winning of bidding on the information bit full flow to be released, especially when
When party in request's platform fails to obtain corresponding knock-down price because bidding failure, remain to according to the knock-down price-bid and win
Rate relation curve predicts the corresponding rate of winning of bidding, and then for directive significance is played in follow-up bid.
Further, the knock-down price-bid wins rate relation curve as shown in Fig. 2 wherein, and the knock-down price can be with
The consumed cost (cost per mile, abbreviation cpm) of every thousand displayings is unit, and the rate of winning of bidding can be with [0,1]
Represent for interval numeral (wherein 0 represents described in the rate of winning of bidding be 0%, 1 represents the rate of winning of bidding as 100%).This area
Technical staff's understanding, the knock-down price-bid and win rate relation curve with incremental characteristic, the incremental property refers to described
The rate of winning of bidding increases with the growth of the knock-down price.
By upper, using the scheme of first embodiment, entered with the corresponding relation of the rate of winning of bidding based on bid than existing
To predict the technical scheme of the rate of winning of bidding, the technical scheme of the embodiment of the present invention can be with the knock-down price to set out for row fitting
Point, the more intuitive and accurate rate of winning of bidding to the information bit to be released is predicted, and is conducive to Nutrition guide needs side's platform
Adjustment bid in time, improves the number of times of winning to the information bit to be released.
Fig. 3 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the second embodiment of the present invention
Figure.Specifically, in the present embodiment, step S201 is first carried out, for the information bit to be released, at least one known to statistics
The bid for representing request and corresponding bid results in bid section.More specifically, known at least one bid
Representing in section asks what can be sent including party in request's platform itself to represent request, can also include and the demand
What its Fang Pingtai associated other party sent represents request.
Performed subsequently into step S202, the bid for representing request and correspondence in described at least one bid section
Bid results, fitting obtain it is described bid and the rate of winning of bidding between bid-rate relation curve of winning of bidding.Preferably,
Bid-rate the relation curve of winning of bidding is obtained according to Gaussian function and error function fitting.Those skilled in the art manage
Solution, this step is used for based on the bid for representing request at least one bid section and corresponding bid results, fitting
Obtain it is the multiple bid section in except described at least one bid section in addition to bid and corresponding bid results (even if
Actual being sent to the information bit to be released does not represent request for these bids), so as to obtain the information bit to be released exist
Bid and corresponding bid results under full flow.
Next step S203 is performed, based on the bid-rate of winning of being bidded described in the determination of rate relation curve of winning of bidding
It is 100% bid section.Specifically, the rate of winning of bidding is that 100% bid section is additionally may included in mathematical meaning
The rate of winning of bidding approaches 100% bid section in upper error range, in the application scenarios by taking Internet advertising position as an example
In, the error range is 5%, then the bid section that the rate of winning is 100% of being bidded described in this step can include described bidding
The rate of winning is 95% to 100% bid section.It will be appreciated by those skilled in the art that party in request's platform is to described to be released
Information bit is sent when representing request, due to the limitation of the factors such as cost, typically will not go to bid with bid higher, then for institute
The bid section that the rate of winning of bidding is 100% is stated, it is difficult to obtain its corresponding bid and knock-down price, then the embodiment of the present invention is based on
The bid for representing request of at least one bid section that party in request's platform described in early stage is actually sent out and corresponding bid results,
Fitting obtains bid-bid win rate relation curve of the information bit to be released on full flow such that it is able to which it is right to predict
Bid section when the rate of winning of being bidded for the information bit to be released is 100%, is conducive to party in request's platform more smart
The bid ranges that accurate determination is adapted with the information bit to be released.
Performed subsequently into step S204, with bid section that the rate of winning of bidding is 100% as the upper limit, by default
The random bid detection of number of times, to obtain the distribution of bid win rate of the information bit to be released on full flow.Specifically,
The preset times pre-set acquisition based on party in request's platform.More specifically, treated described in the distribution of the rate of winning of bidding
Release news the bid results on full flow.Preferably, the bid results include that the random bid of the preset times is visited
In survey, the number of times of winning of each bid, and the corresponding knock-down price of bidding when winning.In a preference, the need
The side's of asking platform is detected by the random bid of 100 times daily, is obtained on the multiple bid section to the information bit to be released
The bid results for representing request, the knock-down price and the knock-down price according to the information bit to be released when winning every time are in institute
The number of times of winning in 100 random bid detections is stated, bid win rate of the information bit to be released on full flow is obtained
Distribution.Preferably, the preset times can be 100 times/day, and those skilled in the art can according to actual needs become and dissolve more
Many embodiments, will not be described here.
Next step S205 is performed, according to the conclusion of the business of the information bit to be released when being won every time in the bid results
Valency, determines the knock-down price-rate relation curve of winning of bidding of the knock-down price and the rate of winning of bidding.Specifically, people in the art
Member may be referred to step S102 described in above-mentioned embodiment illustrated in fig. 1, will not be described here.
Step S206 is finally performed, based on the knock-down price-the rate relation curve prediction of winning of bidding bids the rate of winning.Specifically
Ground, those skilled in the art may be referred to step S103 described in above-mentioned embodiment illustrated in fig. 1, will not be described here.
By upper, using the scheme of second embodiment, the step S201, the step S202, the step S203 and
The step S204 can be understood as a specific embodiment, this reality of step S101 described in above-mentioned embodiment illustrated in fig. 1
The example technical scheme is applied in order to understand the influence that different bid sections may be caused to the rate of winning of bidding, is preferably carried out pre-
If the bid detection of number of times, is especially bid-bidded the rate of winning as 100% of being bidded determined by rate relation curve of winning with early stage
Bid section on the basis of the upper limit, high price detection to be carried out, to obtain information bit to be released the bidding on full flow
As a result, so as to provide more comprehensive fitting sample to fit the knock-down price-bid rate relation curve of winning.
Fig. 4 is the flow of the Forecasting Methodology of the rate of winning of bidding of the information bit a kind of to be released of the second embodiment of the present invention
Figure.Specifically, in the present embodiment, step S301 is first carried out, for the information bit to be released, at least one known to statistics
The bid for representing request and corresponding bid results in bid section.More specifically, those skilled in the art can join
It is admitted to and states step S201 described in embodiment illustrated in fig. 3, will not be described here.
Performed subsequently into step S302, the bid for representing request and correspondence in described at least one bid section
Bid results, fitting obtain it is described bid and the rate of winning of bidding between bid-rate relation curve of winning of bidding.Specifically,
Those skilled in the art may be referred to step S202 described in above-mentioned embodiment illustrated in fig. 3, will not be described here.
Next step S303 is performed, based on the bid-rate of winning of being bidded described in the determination of rate relation curve of winning of bidding
It is 100% bid section.Specifically, those skilled in the art may be referred to step described in above-mentioned embodiment illustrated in fig. 3
S203, will not be described here.
Performed subsequently into step S304, with bid section that the rate of winning of bidding is 100% as the upper limit, by default
The random bid detection of number of times, to obtain the distribution of bid win rate of the information bit to be released on full flow.Specifically,
Those skilled in the art may be referred to step S203 described in above-mentioned embodiment illustrated in fig. 3, will not be described here.At one preferably
In example, the bid and corresponding random probing probability detected for random bid can be calculated with offline mode, so as to need
To be sent to the information bit to be released when representing request, be randomly selected by rand random numbers and foregoing calculated with offline mode
The bid of the random bid detection for obtaining is bid, and with based on normal flow (for example, information retrieval to be released is sorted
Deng) information to be released that determines send together it is described represent request, so as to complete this be bidded to the information bit to be released
Operation.
Next step S305 is performed, according to the result of the random bid detection, obtains described pending when winning every time
The knock-down price of cloth information bit and corresponding number of times of winning.Specifically, the result of the random bid detection includes going out each time
Valency detects corresponding bid results.
The result of the random bid detection of the preset times of table 1
| Knock-down price (is divided) | Win number of times |
| 5 | 2 |
| 10 | 5 |
| 15 | 5 |
| … | … |
| 100 | 1 |
| 200 | 1 |
Table 1 shows that party in request's platform counts what is obtained after having carried out 100 random bid detections altogether, wins every time
The knock-down price of Shi Suoshu information bits to be released and corresponding number of times of winning, wherein, the knock-down price with represent for every thousand times into
This is unit, the monetary unit of the cost cpm for representing for described every thousand times can for point.
Performed subsequently into step S306, for each knock-down price, calculated based on the corresponding number of times of winning of the knock-down price
Obtain the rate of winning of bidding of the knock-down price.
Table 2 calculates the corresponding rate of winning of bidding of different knock-down prices for obtaining according to table 1
| Knock-down price (is divided) | Bid the rate of winning |
| 5 | 2% |
| 10 | 7% |
| 15 | 12% |
| … | … |
| 100 | 99% |
| 200 | 100% |
Table 2 shows that the computational methods calculate the corresponding victory of bidding of the knock-down price for obtaining according to embodiments of the present invention
Extracting rate, specific computational methods those skilled in the art may be referred to the description in above-mentioned embodiment illustrated in fig. 1, not superfluous herein
State.In the application scenarios by taking Internet advertising as an example, for the bid results that knock-down price is 15 points, its corresponding bidding is won
Rate is (5+5+2)/100=12%.Preferably, the rate of winning of bidding can be represented with percents, it is also possible to [0,1]
Numerical value represent that those skilled in the art can according to actual needs become and dissolve more embodiments, and this has no effect on of the invention
Technology contents.
Step S307 is finally performed, determines to obtain the conclusion of the business according to the knock-down price and corresponding rate of winning of bidding
Valency-rate relation curve of winning of bidding.Preferably, the knock-down price and bidding is won according to Gaussian function and error function
Rate is processed, and the knock-down price-rate relation curve of winning of bidding is obtained to be fitted.In a preference, based on following public affairs
Formula is calculated and obtains the knock-down price-rate relation curve of winning of bidding：
F (x)=ω1(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+erf(x))
Wherein, f (x) is the knock-down price-rate relation curve, ω of winning of biddingiIt is i-th weight of Gaussian function point
Cloth, x is the rate of winning of bidding, and erf (x) is error function, and n is default Gauss curve fitting quantity, 1≤i≤n.
Further, the error function is based on equation below and calculates acquisition：
Wherein, erf (x) is the error function, and x is the rate of winning of bidding, and t is Gaussian parameter；The Gaussian parameter is based on
Formula t=(x-m)/(2 × δ) is calculated and obtained, wherein, x is the rate of winning of bidding, and m is equal for the normal distribution of Gaussian function
Value, δ is the covariance of the normal distribution of Gaussian function.
It will be appreciated by those skilled in the art that because the Gaussian function is normal distyribution function, it does not have incremental property；And
The error function is accumulation curve, and it can be understood as the integration for Gaussian Profile (alternatively referred to as normal distribution), quite
In the area that the Gaussian Profile is calculated on the X-Y scheme constituted in the Gaussian Profile, so that by the error letter
Several after-treatments more meets actual knock-down price-rate relation curve of winning of bidding to obtain so that the knock-down price-victory of bidding
Extracting rate relation curve can have incremental characteristic.
In a preference, (corresponding Gaussian function can turn into three Gausses to the default Gauss curve fitting quantity n=3
Function), then won the binary array that rate difference constitutes by being input into the Gaussian function knock-down price and bidding, can obtain
Corresponding output result is obtained, the output result includes three group of three Gaussian function parameter ω1、ω2And ω3；m1、m2And m3；δ1、δ2
And δ3, wherein ω1、ω2And ω3Three weights for representing three Gaussian function；m1、m2And m3For representing that described three is high
Three averages of the normal distribution of this function；δ1、δ2And δ3Covariance for representing the normal distribution of three Gaussian function,
It is fitted in the formula of the optional numerical value substitution error function from three group of three Gaussian function parameter every time, is led to
Cross the fitted data selected from three group of three Gaussian function parameter of adjustment and close obtaining different knock-down price-rates of winning of bidding
It is curve, so as to obtain as shown in Figure 2 best suit actual knock-down price-rate relation curve of winning of bidding.
Preferably, described bidding wins rate difference for reflecting in the bid results, and different cost prices are corresponding to bid
The size of the growth rate (or rate of descent) of rate of winning, for example, the rate difference=cost price of winning of bidding is bidded for 200 timesharing are corresponding
The corresponding rate of winning of the bidding when rate of winning-cost price is 100.
Step S308 is finally performed, based on the knock-down price-the rate relation curve prediction of winning of bidding bids the rate of winning.Specifically
Ground, those skilled in the art may be referred to step S103 described in above-mentioned embodiment illustrated in fig. 1, will not be described here.
By upper, using the scheme of 3rd embodiment, the step S305, the step S306, the step S307 can be with
It is interpreted as one of step S205 described in step S102 described in above-mentioned embodiment illustrated in fig. 1 or above-mentioned embodiment illustrated in fig. 2
Individual specific embodiment, the knock-down price and the corresponding rate of winning of bidding obtained to detection by Gaussian function and error function
The fitting sample of composition is processed, to fit more accurate, intuitively knock-down price-rate relation curve of winning of bidding so that
The curve of acquisition is more smooth while incremental property is kept, and is conducive to the follow-up prediction to the rate of winning of bidding.
Fig. 5 is the structure of the prediction meanss of the rate of winning of bidding of the information bit a kind of to be released of the fourth embodiment of the present invention
Schematic diagram.It will be appreciated by those skilled in the art that prediction meanss 4 described in the present embodiment are used to implement above-mentioned Fig. 1 to embodiment illustrated in fig. 4
Described in method and technology scheme.Specifically, in the present embodiment, the prediction meanss 4 include acquisition module 41, for obtaining
Bid results of the information bit to be released on full flow, the full flow includes the multiple for the information bit to be released
Represent request, it is each to represent request and include corresponding bid, it is the multiple to represent the multiple sections of bidding of request covering；Determining module
42, for according in the bid results every time win when the information bit to be released knock-down price, determine the knock-down price with
The knock-down price of the rate of winning of bidding-rate relation curve of winning of bidding；And prediction module 43, for based on the knock-down price-bid
Rate of winning relation curve predicts the rate of winning of bidding.
Preferably, the acquisition module 41 includes statistic submodule 411, and for the information bit to be released, statistics is known
At least one bid section in the bid for representing request and corresponding bid results；First fitting submodule 412, for root
According to described at least one bid section in the bid for representing request and corresponding bid results, fitting obtain it is described bid with it is competing
Bid-rate relation curve of winning of bidding that valency is won between rate；Determination sub-module 413, for based on the bid-bid and win
Rate relation curve is bidded the bid section that the rate of winning is 100% described in determining；And first treatment submodule 414, for institute
It is the upper limit to state the bid section that the rate of winning of bidding is 100%, is detected by the random bid of preset times, described pending to obtain
The distribution of bid win rate of the cloth information bit on full flow.
Preferably, the first fitting submodule 412 includes fitting unit 4121, for according to Gaussian function and error
Function Fitting obtains the bid-rate relation curve of winning of bidding.
Preferably, the determining module 42 includes acquisition submodule 421, for the knot according to the random bid detection
Really, the knock-down price and corresponding number of times of winning of the information bit to be released when winning every time are obtained；Calculating sub module 422 is right
In each knock-down price, the rate of winning of bidding for obtaining the knock-down price is calculated based on the corresponding number of times of winning of the knock-down price；Second
Treatment submodule 423, determines to obtain the knock-down price-rate of winning of bidding according to the knock-down price and corresponding rate of winning of bidding
Relation curve.
Further, the determining module 42 also includes the second fitting submodule 424, for according to Gaussian function and mistake
Difference function fitting obtains the knock-down price-rate relation curve of winning of bidding.
Preferably, the second fitting submodule 424 includes the first computing unit 4241, for being calculated based on equation below
Obtain the knock-down price-rate relation curve of winning of bidding：
F (x)=ω1(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+erf(x))
Wherein, f (x) is the knock-down price-rate relation curve of winning of bidding, and ω i are i-th weight of Gaussian function point
Cloth, x is the rate of winning of bidding, and erf (x) is error function, and n is default Gauss curve fitting quantity, 1≤i≤n.
Preferably, the second fitting submodule 424 also includes the second computing unit 4242, based on based on equation below
Calculate and obtain the error function：
Wherein, erf (x) is the error function, and x is the rate of winning of bidding, and t is Gaussian parameter；And the 3rd computing unit
4243, the Gaussian parameter is obtained for being calculated based on formula t=(x-m)/(2 × δ), wherein, x is the bid rate of winning, m
It is the average of the normal distribution of Gaussian function, δ is the covariance of the normal distribution of Gaussian function.
More contents of operation principle, working method on the prediction meanss 4, are referred to the phase in Fig. 1 to Fig. 4
Description is closed, is repeated no more here.
One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment is can
Completed with instructing the hardware of correlation by program, the program can be stored in computer-readable recording medium, to store
Medium can include：ROM, RAM, disk or CD etc..
Although present disclosure is as above, the present invention is not limited to this.Any those skilled in the art, are not departing from this
In the spirit and scope of invention, can make various changes or modifications, therefore protection scope of the present invention should be with claim institute
The scope of restriction is defined.
Claims (14)
1. a kind of Forecasting Methodology of the rate of winning of bidding of information bit to be released, it is characterised in that comprise the following steps：
Bid results of the information bit to be released on full flow are obtained, the full flow includes being directed to the information to be released
The multiple of position represents request, each to represent request and include corresponding bid, and the multiple request covering multiple that represents is bid sections；
According to the knock-down price of the information bit to be released when being won every time in the bid results, determine the knock-down price and bid
The knock-down price of rate of winning-rate relation curve of winning of bidding；
Based on the knock-down price-the rate relation curve prediction of winning of bidding bids the rate of winning.
2. Forecasting Methodology according to claim 1, it is characterised in that obtain the information bit to be released on full flow
Bid results, comprise the following steps：
For the information bit to be released, statistics known at least is bid the bid for representing request in section and corresponding
Bid results；
The bid for representing request and corresponding bid results in described at least one bid section, are fitted described in obtaining
Bid-rate relation curve of winning of bidding between valency and the rate of winning of bidding；
Based on the bid-the bid section that the rate of winning is 100% of being bidded described in the determination of rate relation curve of winning of bidding；
With bid section that the rate of winning of bidding is 100% as the upper limit, detected by the random bid of preset times, to obtain
The distribution of bid win rate of the information bit to be released on full flow.
3. Forecasting Methodology according to claim 2, it is characterised in that the bid-bid rate relation curve of winning is based on
Following steps fitting is obtained：
Bid-rate the relation curve of winning of bidding is obtained according to Gaussian function and error function fitting.
4. Forecasting Methodology according to claim 2, it is characterised in that according to when being won every time in the bid results
The knock-down price of information bit to be released determines the knock-down price-rate relation curve of winning of bidding of the knock-down price and the rate of winning of bidding, bag
Include following steps：
According to the result of the random bid detection, the knock-down price and correspondence of the information bit to be released when winning every time are obtained
Number of times of winning；
For each knock-down price, acquisition the bidding for knock-down price is calculated based on the corresponding number of times of winning of the knock-down price and is won
Rate；
Determine to obtain the knock-down price-rate relation curve of winning of bidding according to the knock-down price and corresponding rate of winning of bidding.
5. the Forecasting Methodology according to any one of Claims 1-4, it is characterised in that the knock-down price-rate pass of winning of bidding
It is that curve is determined based on following steps：
Knock-down price-rate the relation curve of winning of bidding is obtained according to Gaussian function and error function fitting.
6. Forecasting Methodology according to claim 5, it is characterised in that described based on Gaussian function and error function fitting
The knock-down price-rate relation curve of winning of bidding is obtained, is comprised the following steps：
Calculated based on equation below and obtain the knock-down price-rate relation curve of winning of bidding：
F (x)=ω1(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+erf(x))
Wherein, f (x) is the knock-down price-rate relation curve, ω of winning of biddingiIt is i-th weight distribution of Gaussian function, x is
Bid the rate of winning, erf (x) is error function, n is default Gauss curve fitting quantity, 1≤i≤n.
7. Forecasting Methodology according to claim 6, it is characterised in that the error function is based on equation below and calculates to obtain
：
Wherein, erf (x) is the error function, and x is the rate of winning of bidding, and t is Gaussian parameter；
The Gaussian parameter is based on formula t=(x-m)/(2 × δ) and calculates acquisition, wherein, x is the rate of winning of bidding, and m is height
The average of the normal distribution of this function, δ is the covariance of the normal distribution of Gaussian function.
8. a kind of prediction meanss of the rate of winning of bidding of information bit to be released, it is characterised in that including：
Acquisition module, for obtaining bid results of the information bit to be released on full flow, the full flow includes being directed to
The multiple of the information bit to be released represents request, each to represent request and include corresponding bid, the multiple to represent request and cover
The multiple bid sections of lid；
Determining module, for the knock-down price according to the information bit to be released when being won every time in the bid results, determines institute
State the knock-down price-rate relation curve of winning of bidding of knock-down price and the rate of winning of bidding；
Prediction module, for based on the knock-down price-the rate relation curve prediction of winning of bidding bids the rate of winning.
9. prediction meanss according to claim 8, it is characterised in that the acquisition module includes：
Statistic submodule, for the information bit to be released, going out for request is represented at least one bid section known to statistics
Valency and corresponding bid results；
First fitting submodule, for bidding the bid for representing request in section and corresponding to bid according to described at least one
As a result, fitting obtains the bid-rate relation curve of winning of bidding between the bid and the rate of winning of bidding；
Determination sub-module, for based on the bid-bid win rate relation curve determine described in the rate of winning of bidding be 100%
Bid section；
First treatment submodule, for bid section that the rate of winning of bidding is 100% as the upper limit, by preset times
Random bid detection, to obtain the distribution of bid win rate of the information bit to be released on full flow.
10. prediction meanss according to claim 9, it is characterised in that the first fitting submodule includes：
Fitting unit, for obtaining the bid-rate relation curve of winning of bidding according to Gaussian function and error function fitting.
11. prediction meanss according to claim 9, it is characterised in that the determining module includes：
Acquisition submodule, for the result according to the random bid detection, obtains the information bit to be released when winning every time
Knock-down price and corresponding number of times of winning；
Calculating sub module, for each knock-down price, is calculated based on the corresponding number of times of winning of the knock-down price and obtains the knock-down price
The rate of winning of bidding；
Second processing submodule, determines to obtain the knock-down price-bid according to the knock-down price and corresponding rate of winning of bidding
Rate of winning relation curve.
12. prediction meanss according to any one of claim 8 to 11, it is characterised in that the determining module includes：
Second fitting submodule, for obtaining the knock-down price-rate of winning of bidding according to Gaussian function and error function fitting
Relation curve.
13. prediction meanss according to claim 12, it is characterised in that the second fitting submodule includes：
First computing unit, the knock-down price-rate relation curve of winning of bidding is obtained for being calculated based on equation below：
F (x)=ω1(1+erf(x))+...+ωi(1+erf(x))+...+ωn(1+erf(x))
Wherein, f (x) is the knock-down price-rate relation curve, ω of winning of biddingiIt is i-th weight distribution of Gaussian function, x is
Bid the rate of winning, erf (x) is error function, n is default Gauss curve fitting quantity, 1≤i≤n.
14. prediction meanss according to claim 13, it is characterised in that the second fitting submodule also includes：
Second computing unit, the error function is obtained for being calculated based on equation below：
Wherein, erf (x) is the error function, and x is the rate of winning of bidding, and t is Gaussian parameter；
3rd computing unit, the Gaussian parameter is obtained for being calculated based on formula t=(x-m)/(2 × δ), wherein, x is described
Bid the rate of winning, m is the average of the normal distribution of Gaussian function, δ is the covariance of the normal distribution of Gaussian function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611191944.7A CN106845682A (en) | 2016-12-21 | 2016-12-21 | The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201611191944.7A CN106845682A (en) | 2016-12-21 | 2016-12-21 | The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106845682A true CN106845682A (en) | 2017-06-13 |
Family
ID=59135109
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201611191944.7A Pending CN106845682A (en) | 2016-12-21 | 2016-12-21 | The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106845682A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846696A (en) * | 2018-06-07 | 2018-11-20 | 北京金山安全软件有限公司 | Advertisement space guaranteed price dynamic adjusting method and device, electronic equipment and storage medium |
-
2016
- 2016-12-21 CN CN201611191944.7A patent/CN106845682A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108846696A (en) * | 2018-06-07 | 2018-11-20 | 北京金山安全软件有限公司 | Advertisement space guaranteed price dynamic adjusting method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Martin et al. | Managers’ green investment disclosures and investors’ reaction | |
| AU2003238004B2 (en) | System and method for estimating and optimizing transaction costs | |
| Mahdavi et al. | Fiscal stringency and fiscal sustainability: Panel evidence from the American state and local governments | |
| TWI242724B (en) | Methods and systems for optimizing return and present value | |
| US20050187851A1 (en) | Financial portfolio management and analysis system and method | |
| CN106910129B (en) | Asset data processing method, client and server | |
| CN109345372A (en) | Credit-graded approach, system and computer readable storage medium | |
| CN103716351A (en) | Information display method and server | |
| CN107093081A (en) | Service strategy formulating method and device | |
| CN110033372A (en) | A kind of method, system and equipment optimizing transaction cost | |
| CN106339904A (en) | Cost Per Sale low-cost marketing promotion method based on electronic commerce | |
| Demery | Côte d’Ivoire: fettered adjustment | |
| Jongwanich et al. | Trade protection and firm productivity: evidence from Thai manufacturing | |
| Frey et al. | The impact of iceberg orders in limit order books | |
| CN107358531A (en) | premium settlement method and device | |
| CN109242313A (en) | Intelligentized innovation item investment value analysis and assessment system | |
| Hunt et al. | Valuation of technology: exploring a practical hybrid model | |
| CN106845682A (en) | The Forecasting Methodology and device of a kind of rate of winning of bidding of information bit to be released | |
| CN115760363A (en) | Interest rate measuring and calculating method and device based on pedestrian credit report | |
| CN106503871A (en) | A kind of stock last-period forecast method of statistics of being voted based on how similar stock | |
| Tondapu | Efficient Market Dynamics: Unraveling Informational Efficiency in UK Horse Racing Betting Markets Through Betfair's Time Series Analysis | |
| CN108876612A (en) | A kind of preferred method suitable for Money Market Fund trade variety | |
| Zhang et al. | The impact of information disclosure on price fluctuations and housing bubbles: an experimental study | |
| KR20190092817A (en) | System for providing information for investment using performance pattern and method thereof | |
| JP2004118813A (en) | Recommendation system for optional article, and method therefor |
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: 20170613 |