CN102640173A - Method and apparatus for configurable model-independent decomposition of a business metric - Google Patents

Abstract

Methods and apparatuses for decomposing a business metric based on a plurality of activities and a response model are described. In one embodiment, the method accesses the response model and the plurality of activities, the plurality of activities each having a reference and executed value. The method computes a contribution to the business metric based on setting one of the plurality of activities to one of the corresponding reference and executed value and setting the other activities to the value state opposite of that activity. Furthermore, the method computes each of the contributions independent of the response model type.

Description

Commerce is measured the method and apparatus that carries out the irrelevant decomposition of configurable model

Technical field

The present invention relates generally to analysis, and more particularly, relate to the influence of confirming that marketing activity is measured commerce multidimensional data.

Background technology

Commercial target is to explain the sales volume result, how to influence sales volume so that understand marketing activity, and for example how many amounts these marketing activities have contributed to total sales volume.The contribution of these trading volumes is useful to the measure of effectiveness (sales volume that for example every cost is a dollar or rate of return on investment (ROI)) that calculates said activity.These some movable instances be directly receive business control activity (for example; Our TV advertisement, to our the representing of product, our rise in price of product), the activity that receives other business control on the market (for example; And/or environment itself (for example cold wave, gas price rise etc.) TV advertisement of the representing of rival, rival etc.).For instance, a certain company possibly want based on these movable or these movable variations know following sell roughly change, the profit of growth and product or service.In addition, many companies want to know that these movable variations (for example, marketing, advertisement, price variation etc.) are to depending on the influence of these movable forecast data (for example, sales volume, growth, profit etc.).

Except that being applied to sales volume, these are constructed also to be used for the decipher activity and can to measure the influence that commerce is measured (for example, income, profit or the market share etc.) to other.

Usually, the analyst uses mathematical model to estimate how these activities are influencing sales volume in the past or in the future.Instance is based on the model of recurrence.Based on the model that returns usually with make in sales volume and the activity each relevant via coefficient.The analyst confirms coefficient based on regression model (or another multivariate technique).The analyst follows the said coefficient of decipher, thinks each movable trading volume rate of change of assigning.For instance, the analyst will confirm that a unit of activity (for example sales promotion) will equal percent X of sales volume or the variation of Y unit.The analyst then multiply by this coefficient the variation of movable amount, and calculates the trading volume of corresponding amount.Through each of activity/coefficient centering is carried out this analysis, analyst's prediction and explanation activity are to the influence of trading volume (or another relevantly can measure commerce and measure, as profit etc.).In addition, change in order to explain the trading volume on cycle some time, the analyst is with the trading volume contribution of computational activity in each time cycle, and report discrepancy, as the explanation that trading volume is changed.

The problem of the method is, employed model is depended in the decipher of the coefficient that draws.For instance, will differ greatly each other to same trading volume and movable cost coefficient in linear model and the multivariate model.This product or gathering on the channel of make using dissimilar models is difficulty, and need employed model form proprietary trading volume decipher algorithm.In addition, the trading volume rate of change depends on selected active set.In addition, when using activity on the same group to the same time cycle, the result is inconsistent.In addition, some activities do not have the trading volume contribution calculation based on natural reference value (for example price, distribution), and therefore, be difficult to confirm these activities in cycle some time to the influence of trading volume with the trading volume variation.

Summary of the invention

The present invention describes and is used for decomposing the method and apparatus that commerce is measured based on a plurality of activities and a response model.In one embodiment, the said response model of said method access and said a plurality of activity, said a plurality of activities have reference value and executed value separately.Said method is set to the state of value opposite with said activity based on one and other activity that one in said a plurality of activities are set in corresponding reference and the executed value, calculates the contribution that said commerce is measured.In addition, said method is independent of the response model type and calculates each in the contribution.

Description of drawings

Unrestricted mode is explained the present invention with instance in the figure of accompanying drawing, wherein same reference indication like.

Fig. 1 is the processing block diagram of the cubical embodiment of explanation trading volume.

Fig. 2 is the processing block diagram of an embodiment of explanation model structure.

Fig. 3 is the table of an embodiment of explanation trading volume decomposition.

Fig. 4 is the process flow diagram that is used to calculate an embodiment who comprises the process that trading volume that synergy distributes decomposes.

Fig. 5 is the table of an embodiment of explanation trading volume decomposition computation.

Fig. 6 is the process flow diagram that is used to calculate an embodiment of the process that synergy distributes.

Fig. 7 is the table of an embodiment of explanation synergy calculating.

Fig. 8 AB is the block diagram that the synergy distribution of carrying out through original value convergent-divergent and absolute value convergent-divergent is described.

Fig. 9 is the processing block diagram that the explanation trading volume is decomposed an embodiment of level.

Figure 10 is used to calculate the process flow diagram that decomposes an embodiment of the process of reporting to the trading volume of atomic Decomposition grade.

Figure 11 is the process flow diagram of an embodiment that is used for confirming the process of atomic Decomposition grade.

Figure 12 is used to calculate the process flow diagram of an embodiment that decomposes the process of report to the trading volume of the concentration range of atomic Decomposition grade.

Figure 13 is used for calculating the process flow diagram that decomposes an embodiment of the process of reporting to the trading volume of the concentration range that decomposes the higher decomposition levels of level.

Figure 14 is the chart of an embodiment of illustration report.

Figure 15 is the block diagram of an embodiment of the difference prediction trading volume in the explanation different time cycle.

Figure 16 is the process flow diagram that is used to calculate a hybrid reason and an embodiment who distributes synergistic process.

Figure 17 is the process flow diagram of an embodiment that is used to calculate the process of compound reason.

Figure 18 is that the calculating trading volume is decomposed report, atomic Decomposition, trading volume decomposition level, the mixing reason is reported and/or the figure of an embodiment of the data handling system that compound reason is reported.

Figure 19 is the figure that is fit to put into practice an embodiment of operating environment of the present invention.

Figure 20 is the figure that is adapted at an embodiment of the data handling systems such as for example general-purpose computing system that Fig. 4, Fig. 6, Figure 10 use in the operating environment of Figure 13, Figure 16 and Figure 17.

Embodiment

In the following detailed description of embodiments of the invention, with reference to accompanying drawing, wherein same reference is indicated like, and wherein shows by means of diagram and can put into practice specific embodiment of the present invention.Describing these embodiment with the details of abundance is in order to make the those skilled in the art can put into practice the present invention; And will understand; Other embodiment be can utilize, and can logic, machinery, electricity, function and other change be made without departing from the scope of the invention.Therefore, be not on limited significance, to carry out following detailed description, and scope of the present invention is only defined by appended claims.

This paper describes and uses response model to come decipher can measure the method and apparatus that commerce is measured.In one embodiment, use response modes to calculate to measure commerce and measure (sales volume, income, profit or the market share etc.).In response model, gather through measurement result and to represent marketing activity.These measurement results are called movable " driver ".The movable driver set of in the model all is called scene.Scene can be represented: the virtual condition in the world, the i.e. actual marketing program of carrying out in true business environment; Or the imaginary state in the world, its reflection is used to plan or the hypothesis to marketing activity and business environment of analysis purpose.

In one embodiment, carrying out decipher commerce through decomposition of calculating trading volume and/or trading volume variance measures.To arbitrary scene or scene (true or imagination) calculated trading volume decomposition and trading volume variance report (being also referred to as " reason report ")." executed " value of terms drive is the value of in scene, being got for driver, and said scene comprises imaginary scene.

In one embodiment, calculate trading volume and decompose, it is irrelevant with the type of response model that is used for sales volume is carried out modeling.Trading volume is decomposed and to be provided the indication to the contribution of sales volume that is produced by marketing activity set.In one embodiment, switch with opening between (on) state at (off) state that closes, calculate trading volume decomposition to said active set through each the driver that is used in the active set.To the off status of activity corresponding to the scene of wherein not carrying out this activity (for example, not carrying out sales promotion or price does not give a discount).In the case, said activity does not increase the trading volume of selling, and takes the driver of reference value to represent by it.The state of opening of said activity increases the contribution to trading volume, and by the executed value representation of the driver of said activity.Difference between the trading volume under the movable open and closed is called as movable original trading volume contribution.In addition, in one embodiment, synergy is distributed to each in the trading volume contribution based on the absolute value of trading volume contribution.

In another embodiment, use based on the trading volume of atomic Decomposition grade and decompose level, calculate trading volume contribution report sequence with different detail grades.The atomic Decomposition grade is one basic " inseparable " or " atom " active set.In addition, define the set of tree level, it is described and how will run up in the gathering activity from the trading volume contribution of these atom actions.In one embodiment, the gathering activity is formed and the level of atomic Decomposition grade in inner consistent trading volume decomposition levels.

In another embodiment, calculate and mix the reason report, mix the reason report to the trading volume variance between two different sales volumes of active set indication.In this embodiment, calculate the trading volume variance through in these activities each is switched between starting value and end value, wherein one in starting value and end value and two the different sales volumes are associated.In addition, use first or second group of activity and variation that the response model that is directed against relative group of activity calculates two basic trading volumes between the different sales volumes.In one embodiment, said active set is not all to need reference value in the driver that is associated each.In addition, in one embodiment, synergy is distributed to each in the trading volume variance based on the absolute value of trading volume variance.

In another embodiment, calculate compound reason report, the trading volume variance of active set between two different sales volumes confirmed in compound reason report.In this embodiment, use difference to decompose the trading volume variance of calculating to a non-distribution active set, said activity has the driver with corresponding reference value separately.The reference value of movable driver representes to be in the activity of off status, and does not add contribution to trading volume.Phrase " movable closing " will be represented to be in its reference value with the movable driver that is associated hereinafter.Mix reason and be used to calculate trading volume variance to a non-distribution active set of the driver that does not have the band reference value.Through from two different sales volumes, one deducting the trading volume variance of calculating to said non-distribution active set, calculate trading volume variance to the distribution activity.In addition, in one embodiment, synergy is distributed to each in the trading volume variance based on the absolute value of trading volume variance.

Trading volume is decomposed

In one embodiment; (hereinafter referred to as the state of " trading volume " is described by measurement result set , and the index of said measurement result is product p, time t and place l in commercial distribution amount that time t measures with appropriate units (unit count, ounce, dollar etc.).Said measurement result set to fixing i is called as driver.Driver is the action that can influence trading volume.The market response model with history (for example, the set of all measurement results, wherein t≤T) is mapped to the trading volume V of time T place to product p and place l _{T, P, l}The market response model can be mapped to the historical trading volume result with the measurement result set that takes place in the past, will predict that maybe the measurement result set is mapped to the prediction trading volume.

In one embodiment, trading volume is expressed as the trading volume data cube, wherein dimension is time, product and place.Fig. 1 is that explanation is to block diagram historical and the cubical embodiment of trading volume that predicts sales volume.In Fig. 1,102 expressions of historical trading volume cube form the data of the time series of multi-dimension data cube, the for example V of preceding text _{T, p, l}Although in one embodiment, the dimension of historical trading volume cube 102 is time, product and place, and that alternate embodiment can have is more, still less and/or different dimensions.Historical trading volume 102 finishes at special time 112 places.Cubical part in actual history trading volume 102 left sides is represented got trading volume the earliest.In addition, in Fig. 1, response model 110A is mapped to historical trading volume 102 with historical act 104.Activity is to have the action of influence to trading volume, and can comprise one or more drivers, further describes like hereinafter.

Yet historical trading volume 102 is not at the appointed time 108 places end always with historical act 104.In other embodiments, historical trading volume 102 is the time cycles to arbitrary past with historical act 104, and has different length, for example some skies, some weeks, some moons, several years etc.In addition, historical trading volume information 102 can have different time length or represent the overlapping time cycle with historical act 104.

In addition, response model 110B is mapped to prediction trading volume 108 with foresight activity 106.In one embodiment, prediction trading volume 108 has identical dimensional with historical trading volume: time, product and place.Foresight activity 106 is to duplicate, derive, derive, import generation or its combination from the user from a certain other products activity from historical act 104 from historical act 104.This embodiment is intended to become the explanation to foresight activity 106, and do not hint foresight activity 106 always present times 108 place begin.Other embodiment of foresight activity 106 can be directed against the time cycle in arbitrary future, and has different length, for example some skies, some weeks, some moons, several years etc.In addition, actual activity 104 can have different time length with foresight activity 106.In one embodiment, response model 110B and response model 110A are identical or different.

The analyst uses response model to estimate movable influence to trading volume.In one embodiment, come describing activity through the driver that can have reference value and executed value separately.The reference value of movable driver representes to be in the activity of " pass " state, means that said activity is to trading volume interpolation contribution.Therefore some movable drivers do not have significant off status, and do not have reference value (for example product is distributed in the number, price in shop wherein etc.).The executed value of movable driver is the value that increases the positive or passive contribution of trading volume.This expression activity is in " opening " state.Movable characterize by driver group and by the scope that influenced by said activity of said driver.

In one embodiment, with the sales volume of commerce be expressed as basic trading volume (for example, the reference price of no sales promotion, all products, medial temperature, no advertisement etc.) with because of carrying out the stack of the incidental transaction amount that said active set produces.These activities be commercial activity (for example; The TV advertisement, to the rise in price of the representing of product, product), other the commercial activity on the market (for example; And/or environment itself (for example cold wave, gas price rise etc.) TV advertisement of the representing of rival, rival etc.).Activity is directed against some combination of product, place and time cycle by some drivers in the driver and characterizes with departing from of its reference value.In this embodiment, therefore come describing activity through a driver set and a scope.

In one embodiment, response model is expressed as the mathematical function of basic trading volume and the trading volume that produces because of active set, as illustrated in the equality (1).

$\mathrm{Volume}={\mathrm{Volume}}_{\mathrm{Base}}+\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\β}}_i\·f_i\left({\mathrm{Activity}}_i\right)---\left(1\right)$

Wherein Volume is historical or the prediction trading volume, Volume _BaseBe basic trading volume, β _iBe Activity _iCoefficient, f _iBe to be applied to Activity _iFunction, and Activity _iBe the activity that influences Volume, for example TV advertisement, product represent, rise in price etc.Though function f in one embodiment, _iBe linear function, but in alternate embodiment, function f _iBe another known in this technology function (for example, logarithmic function, exponential function, algebraic function etc.) and/or its combination.For instance; In one embodiment; In the multiplication model, use natural logarithm to represent constant elastic model; Use normalization with the merging on product and the place with shrink modeling, using advertisement retention constant proportion is the lagging influence modeling of marketing action to behavior (for example TV), and to use saturation degree be diminishing returns (or fatigue) modeling of marketing action to behavior.In addition, in one embodiment, each activity is modeled as the function of one or more drivers

Activity _i＝g _i(d ₁，d ₂，…，d _n) (2)

Activity wherein _iBe the activity that influences Volume, g _iBe with driver (d _l, d ₂..., d _n) be transformed to Activity _iFunction, (d ₁, d ₂..., d _n) for influencing the driver of Volume.For instance; In one embodiment; The activity price comprises driver NoPromoPrice, in a given week desired price during no sales promotion, and AvgNoPromoPrice, AvgNoPromoPrice are the average prices of the product under the situation of no sales promotion, sold in given 1 year.As another instance, in one embodiment, movable marketing comprises television advertising (TV).In addition, in one embodiment, as mentioned below driver is used for one or more activities.

Though in one embodiment, use equality (1) and (2) to calculate trading volume, in alternate embodiment, equality (1) and (2) are used for other purpose (scene analysis, forecast, judgement generation, financial prediction etc.).Equality (1) and (2) comprise an embodiment of response model.In addition, response model can have and equality (1) and (2) various embodiment.For instance, the alternate embodiment of response model is general parameterized form (equality (3)):

$\mathrm{Volume}=\hat{f}(A_1,A_2,\·\·\·,A_n,\stackrel{\→}{\mathrm{\β}})---\left(3\right)$

Wherein Volume is historical or the prediction trading volume,

Be model form, A ₁, A ₂..., A _nBe active set, and

It is the vector of coefficient.This comprises gets movable tolerance as any calculating of importing and returning the numeral of the tolerance of describing sales volume.

In addition, response model is expressed as rate pattern, equality (4):

$\frac{\mathrm{Volume}}{\mathrm{ACV}}=\tilde{f}(A_1,A_2,\·\·\·,A_n)---\left(4\right)$

Wherein Volume is historical or the prediction trading volume,

Be model form, A ₁, A ₂..., A _nBe active set, and ACV is all commodity transaction amounts, it is the tolerance of the scale in given place.Also can response model be modeled as the sales promotion conditions model, as in figure (5) and (6):

$\mathrm{Volume}=\underset{\mathrm{PromoCond}}{\mathrm{\Σ}}{\mathrm{Volume}}_{\mathrm{PromoCond}},$

PromoCond ∈ characteristic, represent, characteristic represents, TPR, NoPromo] (5)

Volume _PromoCond＝ACV _PromoCond·f _PromoCond(A ₁，A ₂，…，A _n) (6)

Wherein Volume is historical or the prediction trading volume, Volume _PromoCondBe the trading volume to sales promotion conditions, f is a model form, A ₁, A ₂..., A _nBe active set, ACV _PromoCondFor having all commodity transaction amounts in the place of specifying sales promotion conditions, and PromoCond for by characteristic, represent, characteristic+represent, temporary reduction (TPR) and/or do not have the type of the sales promotion conditions that sales promotion forms.

Fig. 2 is the processing block diagram of an embodiment of explanation response model structure.Though in one embodiment; Fig. 2 is the universal architecture of response model; But in alternate embodiment, the response model structure is model structure known in this technology (for example, parameterized form, rate pattern, sales promotion conditions model, neural network model, based on agency's model etc.).In Fig. 2, model 202 comprise coefficient 204A to 204F, variable 206A to 206F, function 208A to 208B and driver 210A to 210H.In one embodiment, variable representes to influence one in the said active set of institute's modeling trading volume.In another embodiment, variable is represented a plurality of activities, or an activity can be represented by a plurality of variablees.In one embodiment, model 202 is wherein variable 206A to be made up to calculate the sequence of calculation of trading volume to 204F to 206F and coefficient 204A.Variable 206A to 206F by driver 210A to 210H and use said driver 210A to form to 208B to the function 208A of 210H.For instance, variable 206A comprises the function 208A of driver 210A and driver 210B.In addition, variable 208B comprises the function 208B of driver 210D and driver 210C.In addition, variable 208C comprises driver 210C and 210E.Variable 208D comprises the function 208B of driver 210F and driver 210C.Variable 206F comprises driver 210G and 210H.In addition, variable need not comprise driver.For instance, variable 206E does not depend on any driver.In one embodiment, such variable is a constant.

In one embodiment, each driver has the scope that particular drive (measurement result) gets into model for this reason, and it is the set in product and place.In one embodiment, the default scoping that driver gets into model is the scope of model itself, and for example the price of the product A in place 1 is the part of model that is used for the product A in place 1.In another embodiment, same driver also can be used for the product B in place 1 in the model to the product A in the place 1, with the influence to product A of the price of describing product B.In addition, range of variables is based on the scope as the driver of the part of said variable.

The variable in the model and the structure of driver have reflected that model is through designing with the activity for history will be considered with prediction trading volume modeling.In one embodiment, below be about how in each model, reflecting the hypothesis of marketing activity:

1. each model has the set of given activity interested.

2. movable the set by the driver with the scope that is associated with each driver describes.

3. each driver has single " pass " state in each product/week/place.The off status of driver is also referred to as the reference value of driver.

4. activity can open or close.If movable closing is used to describe this movable All Drives so and all is in its off status.If activity is opened, at least one driver is not in its reference value so.

5. the state of the model that all closes of all activities is a base state.The trading volume that is associated with base state is basic trading volume.

In one embodiment, same driver is used for different variablees (and/or movable) and has different range.For instance, driver 210C is used for variable 206B and 206D.In this example, driver 210C will have a scope to variable 206B, and have another scope to variable 206D.

The instance of the relation between movable (and/or variable) and the individual drive is that movable end cover represents.Movable end cover represents the marketing activity of the end on the island, supermarket in the high flow volume region that expression is placed on product in the shop.In one embodiment, be to have the number percent in the shop that the end cap to product represents and the function of the price (being generally price reduction) that is associated with product on representing with the end cap representation of activity.This movable driver is to have the said shop number percent that represents with promotional price.This movable corresponding off status is zero a shop number percent and corresponding to the price (for example, not having price reduction) of underlying price.

For these defined active sets and the driver that is associated, the data that the described response model of preceding text allows the analyst to calculate basic trading volume and produce because of employed active set in the model.In addition, this model allows the analyst to calculate from each the contribution to trading volume in the said active set.Each activity can have actively trading volume, passive or insignificant influence.For instance, the price to product gives a discount and can have positive impact to trading volume.On the contrary, the price that improves on the said identical product can have negative influence to trading volume.Calculate and to be called as trading volume to each the trading volume contribution in the active set and to decompose.Trading volume is decomposed the permission analyst and is confirmed that which person in the active set makes maximum or minimum trading volume contribution.

Fig. 3 is the table of an embodiment of explanation trading volume decomposition.In Fig. 3, basic trading volume 304 is 89.07% of a total amount of transactions.In this embodiment, movable 306A accounts for 11.42% of total amount of transactions to 306M, and remaining-0.35% owing to model error.Movable 306A each in the 306M is made different contributions to total amount of transactions.For instance, characteristic activity 306A makes maximum contributions with 2.99%, and TV advertisement 306E and temporary reduction (TPR) 306D also makes total amount of transactions and is higher than 2% contribution.On the contrary, some activities are not made contributions to total amount of transactions or are made contribution seldom, for example characteristic with represent 306B, competDistrib8thCont (specific rival's distribution) 306L and competDistribPL (distribution of special tags product) 306M.

In one embodiment, model error 302 is poor between the total amount of transactions that calculated (comprising all movable and distributed synergies) and the real trade amount.Model error 302 is reported as an independent classification, or is contained in basic trading volume, active set and/or its combination.

Fig. 4 is the process flow diagram that is used to calculate an embodiment who comprises the process 400 that trading volume that synergy distributes decomposes.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 400 is carried out by the data handling system 1800 of Figure 18.

Referring to Fig. 4, handling frame 402 places, process begins to carry out the trading volume decomposition computation through processing logic access input.In one embodiment, these inputs comprise response model, said model parameter, to the contributive active set of the trading volume that is calculated and with said activity in the set of each driver that is associated.In one embodiment, input is corresponding to the set of response model, coefficient, function and activity and driver, such as preceding text with respect to Fig. 2 description.

Handling frame 404 places, processing logic is confirmed each the driver reference value in the active set.This is through user's input, carries out based on the rule of historical driver values or based on any other logic of historical data or default value.According to preceding text, the driver reference value is represented the off status of driver.In addition, movable off status means that this activity will not produce the additional contribution to trading volume.The activity that will be in off status is defined as to have and comprises that activity will be in the All Drives of off status.

Handling frame 406 and 408 places, processing logic uses response model to calculate basis and prediction trading volume with movable with these movable references and executed value respectively.In one embodiment, processing logic is set to off status and the driver reference value is used for basic trading volume calculate basic trading volume through in the activity each.According to preceding text, basic trading volume is illustrated in commercial does not carry out the sales volume under the situation of any activity to product represented in the trading volume cube, place and/or time cycle.In addition, processing logic uses the executed driver values to calculate the prediction trading volume.In one embodiment, executed driver values is the driver values that actual use or plan are used.

In one embodiment, through the response model of institute's access, processing logic calculates different scenes that some activity is opened and some activity is closed.The difference that produces owing to the switching activity ON/OFF in the prediction trading volume is owing to those movable original trading volumes of switching.Trading volume owing to active set possibly not be the summation owing to the trading volume of indivedual activities.This is non-linear owing to the trading volume response that exists in the real life and in model, capture.Different activities can " help " or " injury " other activity, and ratio produces more or less trading volume under the situation about together carrying out under situation about carrying out separately.Difference between the prediction trading volume is called synergy, and synergy is through reporting or distribute to indivedual activities separately.Hereinafter further describes at processing frame 420 places and calculates synergy.

Processing logic is carried out cycle of treatment (handling frame 410 to 418), to calculate each movable original trading volume contribution.Handling frame 412 places, processing logic is set to all other movable inverse state with one in the activity.In one embodiment, the driver of one in the processing logic activity is set to its off status, and all other movable drivers are set to out state.In this embodiment, processing logic use subtraction scheme is hereinafter described calculated the original trading volume contribution of said activity.In another embodiment, the driver of one in the processing logic activity is set to out state, and all other activities are set to off status.In this embodiment, processing logic use addition scheme is hereinafter described calculated the original trading volume contribution of said activity.

Handling frame 414 places, processing logic uses one in addition and the subtraction scheme to come the original trading volume contribution of computational activity.Through using the active set { a that is imported ₁, a ₂..., a _n, order

Indicative of active i opens, and makes the driver that is associated with said activity get its executed value.Correspondingly; Represent that through

movable i closes, make the driver that is associated with said activity get its reference value to the scope that is associated.

In one embodiment, processing logic uses the addition scheme to calculate original trading volume contribution.In the addition scheme, it is poor that processing logic calculates wherein the trading volume of the scene that scene that single-unit activity only opens and all activities wherein all close.Said difference is the original trading volume contribution owing to said activity, shown in equality (7):

${\mathrm{Volume}}^{\mathrm{add}}\left(a_i\right)=\mathrm{Volume}\left(\right\{a_1^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_{i-1}^0,a_i^1,a_{i+1}^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^0\left\}\right)-\mathrm{Volume}\left(\right\{a_1^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_{i-1}^0,a_{\mathrm{<figure-callout\; id="0"\; label="i"\; filenames="HPA00001357903300051.png,HPA00001357903300151.png"\; state="\{\{state\}\}">i</figure-callout>}}^0,a_{i+1}^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^0\left\}\right)---\left(7\right).$

In another embodiment, processing logic uses the subtraction scheme to calculate original trading volume contribution.In the subtraction scheme, it is poor with the trading volume of the trading volume of under the situation of single-unit activity pass, being predicted that processing logic calculates in the trading volume under the situation that all activities are opened, shown in equality (8):

${\mathrm{Volume}}^{\mathrm{Subtr}}\left(a_i\right)=\mathrm{Volume}\left(\right\{a_1^1,\&CenterDot;\&CenterDot;\&CenterDot;,a_{i-1}^1,a_{i+1}^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^1\left\}\right)-\mathrm{Volume}\left(\right\{a_1^1,\&CenterDot;\&CenterDot;\&CenterDot;,a_{i-1}^1,a_{i+1}^1,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^1\left\}\right)---\left(8\right).$

Cycle of treatment is being handled the end of frame 418 places.

Handling frame 420 places, processing logic is confirmed the synergy contribution, and a said synergistic part is distributed to each indivedual movable original trading volume contribution.Synergy is by non-linear in the model and produce.In order to confirm synergy, processing logic calculates the summation from the movable original trading volume of handling frame 410 to 418, and calculates the increment trading volume.Through combination are poor between prediction trading volume and the prediction trading volume when all activities are all invalid when all activities are all effective from all movable increment trading volumes.This is irrelevant with the method for contributing so as to the original trading volume of using equality (9) to calculate indivedual activities:

$\mathrm{IncVolume}=\mathrm{Volume}\left(\right\{a_1^1,a_2^1,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^1\left\}\right)-\mathrm{Volume}\left(\right\{a_1^0,a_2^0,\&CenterDot;\&CenterDot;\&CenterDot;,a_n^0\left\}\right)---\left(9\right).$

In one embodiment since the nonadditivity in the world with and model representation, the increment trading volume can be not equal to the summation of original trading volume contribution, equality (10a) and (10b):

$\mathrm{IncVolume}\&NotEqual;\underset{i}{\mathrm{\&Sigma;}}{\mathrm{Volume}}^{\mathrm{add}}\left(a_i\right)---\left(10a\right)$

$\mathrm{IncVolume}\&NotEqual;\underset{i}{\mathrm{\&Sigma;}}{\mathrm{Volume}}^{\mathrm{subtr}}\left(a_i\right)---\left(10b\right)$

Difference between increment trading volume and the contribution of total original trading volume is defined as synergy, equality (11a) and (11b):

${\mathrm{Synergy}}^{\mathrm{add}}=\mathrm{IncVolume}-\underset{i}{\mathrm{\&Sigma;}}{\mathrm{Volume}}^{\mathrm{add}}\left(a_i\right)---\left(11a\right)$

${\mathrm{Synergy}}^{\mathrm{subtr}}=\mathrm{IncVolume}-\underset{i}{\mathrm{\&Sigma;}}{\mathrm{Volume}}^{\mathrm{subtr}}\left(a_i\right)---\left(11b\right).$

Synergy means that some or all of marketing activities all cause " summation is greater than part ".This means that carrying out a plurality of activities (as long as at least these activities trading volume is made positive contribution) simultaneously can produce the trading volume that is higher than (or being lower than) summation through individually carrying out all movable trading volumes that produce.Therefore, the additivity synergy trends towards to positive.For the same reason, subtracting property synergy trends towards to passiveness, has movablely not only lost the lifting from said activity individually because close, and has made all the other activity availabilities low slightly.

In one embodiment, and according to increment trading volume and the relative size of total amount of transactions and the relative lifting of different activities, the selection of decomposing scheme has influence to original increment trading volume.For instance, compare with the additivity method, subtracting property method can be distributed to " less " influence with relative high transaction amount contribution.

In addition, handling frame 420 places, and in one embodiment, processing logic is distributed to each in the original trading volume contribution with the synergy that is calculated.Through synergy being distributed to original trading volume contribution, processing logic can calculate the trading volume contribution with the activity of real trade amount contribution modeling.In one embodiment, processing logic is to each the movable synergy of distributing in the active set.In this embodiment, the synergy that is calculated is all activity results of interaction.In another embodiment, processing logic distributes one or more the eliminating activity from synergy.In this alternate embodiment, distribute the activity of getting rid of to have from synergy and equal the final trading volume contribution that original trading volume is contributed.For instance, if movable, get rid of said activity so, or will will form the basic trading volume of being reported after a while with the activity that basic trading volume makes up with additivity mode (for example, replenishing trading volume) entering response model.

In alternate embodiment, among the activity in covering the synergy distribution, distribute synergy based on different formulas.Instance is that proportional distribution, the proportional distribution of absolute size with the trading volume contribution, the distribution that is moiety or feasible each movable summation synergy part of distributing to equal total synergistic any allocative decision.

Before synergistic distribution was discussed, the instance that provides the total amount of transactions decomposition computation was useful.In one embodiment, processing logic distributes the synergy with following characteristic: original and final trading volume contribution has same sign (no sign orientation change); Final trading volume contribution is as far as possible near original contribution; And the relative size of the approaching as far as possible original contribution of the relative size of final contribution.Hereinafter further describes in Fig. 6 and combines this embodiment to distribute synergy.

Fig. 5 is the table that an embodiment of the trading volume decomposition computation of using subtracting property scheme is described.In Fig. 5, processing logic use activity 524A calculates trading volume to 524F and decomposes 522.Movable 524A to 524F comprise TV advertisement 524A, printed matter advertisement 524B, reward voucher 524C, represent 524D, characteristic 524E and compete 524F.Processing logic uses response model (not shown), said response model use driver 510A to 510J with this marketing amount modeling.Movable 524A each in the 524F comprises driver 510A one or more in the 510J.For instance; TV advertisement 524A comprises the total audience ratings driver of TV 510A; Printed matter advertisement 524B comprises printed matter propagation driver 510H; Reward voucher 524C comprises that reward voucher propagates driver 510F, represents 524D and comprise about the percent basis trading volume driver 510B that represents and show price driven device 510C; Characteristic activity 524E comprises about the basic trading volume number percent driver 510D of characteristic and characteristic price driven device 510E, and competitive activities 524F comprises the trading volume number percent driver 510J about trade war.As illustrated in fig. 5, movable 524A comprises driver 510A one or more in the 510J to 524F.Yet, be not all driver 510A to 510J all be contained in movable 524A in the 524F one in.For instance, the total audience ratings 510G of underlying price driver 510I and radio be not contained in movable 524A in the 524F one in.But these drivers increase basic trading volume 508, and constant during the simulation of contributing to the original trading volume of 524F in order to definite movable 524A.

In addition, as indicated above, driver 510A each in the 510J has reference value and executed value.Processing logic uses the executed value to calculate the expection trading volume, and processing logic uses driver 510A to calculate basic trading volume to the reference value of 510J.Processing logic uses reference or desired value to come the original trading volume contribution of computational activity sometimes.Through using these values, processing logic calculate the expection trading volume be 1000 with basic trading volume be 700.The increment trading volume is 300.

Processing logic uses this model mentioned above to come the original trading volume contribution of computational activity 524A each in the 524F.For instance, for TV advertisement 524A, processing logic calculates the trading volume of using the total audience ratings driver of TV 510A, and it becomes reference value 0 from its desired value 20.This scene provides prediction trading volume 950, means that the original trading volume to TV advertisement 524A is 50.For the movable 524B of printed matter, from 1,000,000 closes 0 to processing logic with printed matter active driver 510H.The original trading volume contribution that this calculating provides the movable 524B of printed matter is 20.Similarly, processing logic uses the reference value of the driver that is associated illustrated in fig. 5 to calculate reward voucher 524C, represents 524D, the original trading volume contribution of characteristic 524E and competition 524F is respectively 10,150,100 and-5.

According to above original trading volume contribution, processing logic calculates the absolute value of original trading volume contribution, and it is respectively 50,20,10,150,100 and 5 to movable 524A to 524F.To describe with respect to Fig. 6 like hereinafter, and synergy that distribute movable to each is directed against movable 524A and is respectively-3.73 ,-1.49 ,-0.75 ,-11.19 ,-7.46 and-0.37 to 524F.This final trading volume contribution that causes to basic trading volume 508 is 700, and movable 524A is respectively 46.27,18.51,9.25,138.81,92.54 and-5.37 to the contribution of 524F.

In trading volume mentioned above was decomposed, processing logic distributed synergy based on the absolute value of original trading volume contribution.Fig. 6 is used for the process flow diagram that absolute value based on original trading volume calculates an embodiment of the process 600 that synergy distributes.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 600 is carried out by the data handling system 1800 of Fig. 8.

In Fig. 6, handling frame 602 places, process is sued for peace through the movable original trading volume contribution of calculating in the processing frame of processing logic to Fig. 4 and is begun.Handling frame 604 places, the synergistic amount of processing logic is set to equal the increment trading volume and deducts original trading volume summation.

Handling frame 606 places, processing logic distributes a synergistic part that is calculated based on the absolute value of original trading volume contribution in the contribution of activity trading volume each.In one embodiment, make V ₁, V ₂..., V _nBy be assigned with those movable original trading volumes contributions of the synergistic part of calculating, and make the S be the synergy of being calculated to be allocated.In one embodiment, use equality (12) to calculate the final trading volume contribution of each movable i:

$V_i^{\mathrm{Final}}=V_i+\frac{\left|V_i\right|}{{\mathrm{\&Sigma;}}_{j=1}^n\left|V_j\right|}\&CenterDot;S---\left(12\right).$

Wherein

Be the final trading volume contribution of movable i, V _iBe the original trading volume contribution of movable i, and S is total synergy of calculating.In addition, because synergy S satisfies inequality

Therefore

And V _iTo have same sign.In addition, the original contribution convergent-divergent with same sign has the factor of relative size, thereby keeps trading volume contribution relative size.

As synergistic instance, Fig. 7 is the table of an embodiment of explanation synergy calculating.Trading volume numeral among Fig. 7 is that expection/basic trading volume and the trading volume contribution from Fig. 5 draws.For instance, increment trading volume 706 has value 300, and it is poor for expection trading volume 704 and basic trading volume 508.Original trading volume contribution 758 sued for peace produce sum 325.Difference between this summation and the increment trading volume is 25, and it is synergy 710.In one embodiment, use the equality (7) of preceding text act synergistically and 710 distribute to the contribution of original trading volume.

As indicated above, processing logic distributes synergy based on the absolute value of original trading volume contribution.In alternate embodiment, processing logic distributes synergy based on the original trading volume contribution of reality.Yet, distribute synergy to have shortcoming based on original trading volume contribution.For instance, for the passiveness synergy, distribute the sign-flip that can make movable trading volume contribution based on original trading volume contribution.Sign-flip can become negative trading volume contribution from positive trading volume contribution with activity, or vice versa.Therefore, sign-flip makes activity smudgy to the qualitative contribution that trading volume has.As indicated above, the absolute value of contributing based on original trading volume distributes synergy not have the sign-flip problem.In addition, the distribution of original trading volume synergy can cause a large amount of convergent-divergents to obtain synergy in a small amount.To describe like hereinafter, this can produce because of the original trading volume contribution of opposite sign.Fig. 8 AB is the block diagram that explanation acts synergistically and distributes based on the absolute value that original trading volume is contributed and original trading volume is contributed.In Fig. 8 A, Figure 80 0 comprises the trading volume contribution of two movable 802A to 802B.In this figure, original trading volume contribution 802A is positive, and greater than negative original trading volume contribution.It is to adjust the distribution of original trading volume contribution 802A to 802B in the opposite direction to 804B that synergy is distributed 804A.It is positive that total institute distributes synergy, because | distribution 804A|＞| distribution 804B|.By contrast, in Fig. 8 B, Figure 88 0 comprises that original trading volume contribution 852A distributes 854A to 854B to 852B and synergy.Because total synergy of being distributed is positive; And contribute 852A to distribute synergy to the absolute value of 852B based on original trading volume; Both just are to 854B so synergy is distributed 854A, and distribute 804A to 804B less than the synergy of the correspondence among Fig. 8 A.

Method mentioned above is calculated to the trading volume of single prediction trading volume (for example, to the single product in single-revolution and the single place) and is decomposed.If trading volume set (a plurality of product/weeks/place) needs trading volume to decompose, calculate so to the trading volume of all individual transaction amounts and decompose, and add the trading volume contribution of respective activity.

As indicated above, in one embodiment, calculate the trading volume decomposition through switching the ON/OFF activity, itself and said response model have nothing to do.Though describe decomposable process according to decomposing trading volume, in alternate embodiment, this process can be used for decomposing other measurable commerce and measures (for example, income, profit or the market share etc.).For instance, in one embodiment, processing logic decomposes another measurable commerce to be measured, and distributes synergy, like preceding text described in Fig. 4 and Fig. 6.

Atomic Decomposition and decomposition level

Trading volume mentioned above is decomposed the decomposition under grain size category of the explanation granularity of response model and active set (that is, based on).Yet, the commercial interesting usually decomposition of understanding under the different grain size grade.For instance, the thing that is regarded as the single-unit activity of the purpose that is used for a report (for example, trade promotion) can be regarded as the set of a plurality of activities that are used for another commercial object (for example, represent, characteristic, characteristic and represent and TPR).As another instance, with the single activity that comprises of TV resolve into national TV, local TV, wired, broadcasting, in the daytime, night etc. one or more are movable individually.Yet owing to the synergistic potential distribution under the different grain size grade, the grade of definition of activities will have influence to the trading volume contribution that belongs to active set.This can cause the inconsistency between the different decomposition report.Return referring to the exemplary alive packet, returns of trade can provide with represent, characteristic, characteristic and represent, the different total amount of transactions contribution of summation of TPR activity trading volume contribution.

For fear of these inconsistencies, the basic set that definition " inseparable " or " atom " is movable, and describe how will run up to the tree set in the gathering activity from the trading volume contribution of these atom actions.Each grade of tree is called decomposition levels, and wherein the grade of leaf node is the atomic Decomposition grade.The rate sequence that starts from the atomic Decomposition grade is called the decomposition level.Through being sued for peace, the trading volume contribution of all atom actions (leaf node) below node that activity is associated obtains this movable trading volume contribution.The level of atomic Decomposition grade and higher decomposition levels is called trading volume and decomposes level.

Fig. 9 is the block diagram that the explanation trading volume is decomposed an embodiment of level 900.In Fig. 9, trading volume is decomposed level 900 and is comprised three decomposition levels: general introduction grade 902, detailed levels 904 and atomic level 906.Decompose level 900 and have Three Estate though trading volume is described, in alternate embodiment, trading volume is decomposed level 900 and is had more or less decomposition levels, and it has identical and/or different trading volume decomposition levels.In particular, trading volume decomposition level 900 can have general introduction grade and/or detailed levels more than one.

Atomic Decomposition grade 906 is the lowest class that trading volume is decomposed level 900, and comprises the thinnest activity granularity.Atomic Decomposition grade 906 comprises movable 912A to 912K, and it is %ACV 912A, number of items 912B, characteristic 912C, represents 912D, characteristic and represent 912E, TPR 912F, national TV 912G, local TV 912H, printed matter 912I, radio 912J and FSI 912K.This decomposition levels is served as the basis of detailed decomposition levels 904 and general introduction decomposition levels 902.

Decomposition levels 904 is the trading volume decomposition levels as the gathering of the activity in the atomic Decomposition grade 906 in detail.Decomposition levels 904 comprises movable 910A to 910I in detail, and it is distribution 910A, characteristic 910B, represents 910C, characteristic and represent 910D, TPR 910E, TV 910F, printed matter 910G, radio 910H and FSI 910I.Movable 910A in the decomposition levels 904 is made up of to 912K one or more the movable 912A from atomic Decomposition grade 906 to 910I in detail.For instance, distribution 910A comprises %ACV 912A and number of items 912B.In addition, characteristic 910B, represent 910C, characteristic and represent that 910D, TPR 910E comprise characteristic 912C respectively, represent 912D, characteristic and represent 912E and TPR 912F in each.TV 910F comprises national TV 912G and local TV 912H.

General introduction decomposition levels 902 is the highest decomposition grades in the trading volume hierarchical level, and presents a minimal number movable trading volume decomposition.In one embodiment, the trading volume decomposition carried out to the classification of 908D owing to extensive movable 908A of general introduction decomposition levels 902 expression.In one embodiment, general introduction decomposition levels 902 comprises movable 908A to 908D, and it is distribution 908A, trade 908B, medium 908C and reward voucher 908D.These movable 908A are to the comfortable more fine-grained movable composition of summarizing the decomposition levels 904 of decomposition levels 902 belows of each origin among the 908D.For instance, distribution 908A comprises distribution 910A.Trade 908B comprises characteristic 910B, represents 910C, characteristic and represent 910D and TPR 910E movable.Medium 908C summarizes media ad campaigns, and comprises that TV 910F, printed matter 910G and radio 910H are movable.Reward voucher 908D general introduction reward voucher is movable, and comprises FSI 910I.

Use this trading volume to decompose level 900, the analyst can use high-grade decomposition 902 to obtain which movable 908A is made the variation of trading volume to 908D the general introduction of which contribution.But this analyst or different analyst's degree of depth inquire into detailed 904 or atom 906 trading volumes decompose, to obtain more fine-grained movable contribution.In addition be to build on the atom trading volume to decompose on 902 because trading volume is decomposed, so trading volume decompose in 904 and 906 each each other and decompose 902 consistent with the atom trading volume.Therefore, the analyst can select the trading volume of suitable analyst's needs to decompose granularity.In one embodiment, these trading volume decomposition levels are consistent, because the contribution of the trading volume at each grade place has identical total synergy, and increase identical total amount of transactions contribution.

Figure 10 is the process flow diagram of an embodiment of the trading volume that is used for the calculation of atomic decomposition levels process 1000 of decomposing report.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving) or the two combination.In one embodiment, process 1000 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 10, handling frame 1002 places, process begins through the defined atom action set of processing logic access.According to preceding text, defined atom action set is the inseparable active set of this response model.Hereinafter further describes defined active set in Figure 11.The original trading volume contribution of processing logic each in handling frame 1004 places calculation of atomic activity.In one embodiment, processing logic uses addition or subtraction scheme to calculate original trading volume contribution handling frame 410 to 418 places, like preceding text described in Fig. 4.

Handling frame 1006 places, processing logic uses response model to calculate basic trading volume.In one embodiment, processing logic calculates basic trading volume through all atom actions are closed, described in the processing frame 408 in Fig. 4 mentioned above.Processing logic is being handled frame 1008 places from basis and expection trading volume calculating increment trading volume.In one embodiment, calculate the increment trading volume described in the processing frame 406 among processing logic such as Fig. 4.

Handling frame 1010 places, the final trading volume contribution of each in the activity of processing logic calculation of atomic.In one embodiment, processing logic calculates final trading volume contribution through distributing the synergy that is calculated, described in the processing frame 420 of preceding text in Fig. 4.Processing logic presents basic trading volume and final trading volume contribution at processing frame 1012 places.In one embodiment, processing logic is with graphics mode, to show or to present this data with another program known in this technology.

Use the atomic Decomposition grade that is calculated to define and/or calculate than the high transaction volume decomposition levels.Figure 11 is used for confirming the process flow diagram than an embodiment of the process 1100 of high transaction volume decomposition levels based on atom trading volume decomposition levels.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 1100 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 11, handling frame 1102 places, said process will begin to the active set of its division trading volume contribution through the processing logic access.Processing logic is carried out cycle of treatment (handling frame 1104 to 1110), to calculate each movable original trading volume contribution.Handling frame 1106 places, processing logic definition higher level trading volume is decomposed group.In one embodiment, processing logic is through defining the new incompatible definition of active set than high transaction volume decomposition group group from low active set.Said low active set is that atom action set or trading volume are decomposed the active set that is higher than the decomposition of atom trading volume in the level and is lower than the active set that is just defining.For instance, in Fig. 9, processing logic based on from the movable 912A of atom trading volume decomposition levels 906 to 912K define be used for detailed transaction amount decomposition levels 904 movable 910A to 910I.Processing logic decompose in the driver in the group to defined trading volume each define reference value set.Cycle of treatment is being handled the end of frame 1110 places.

Use defined different trading volume decomposition levels, described in Figure 12 and Figure 13, calculate trading volume decomposition level like hereinafter.Figure 12 is used to calculate the process flow diagram of an embodiment that decomposes the process 1200 of report to the trading volume of the concentration range of atomic Decomposition grade.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 1200 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 12, handling frame 1202 places, said process begins with final trading volume contribution through the basic trading volume of processing logic calculation of atomic trading volume decomposition levels.In one embodiment, processing logic such as Fig. 4 of preceding text description calculate basic trading volume and final trading volume contribution.Processing logic is sued for peace to each movable respective transaction amount contribution on the basis in the inter-product/week/place in the said scope at processing frame 1204 places.Processing logic presents basic trading volume and final trading volume contribution at processing frame 1012 places.In one embodiment, processing logic is with graphics mode, to show or to present this data with another program known in this technology.

Figure 13 is the process flow diagram of an embodiment of the trading volume that is used for calculating the concentration range that decomposes the higher decomposition levels of the level process 1300 of decomposing report.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 400 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 13, handling frame 1302 places, said process begins with final trading volume contribution through the basic trading volume of processing logic calculation of atomic trading volume decomposition levels.In one embodiment, processing logic such as Fig. 4 of preceding text description calculate basic trading volume and final trading volume contribution.

Handling frame 1304 places, decompose group for each trading volume, the trading volume contribution that processing logic decomposes group to the atom trading volume in the leaf node that belongs to the trading volume decomposition group in the level is sued for peace.In one embodiment, each leaf node is the activity of one in the trading volume decomposition levels.In addition, processing logic is assigned this summation to this trading volume and is decomposed group.Processing logic presents basic trading volume and final trading volume contribution at processing frame 1012 places.In one embodiment, processing logic is with graphics mode, to show or to present this data with another program known in this technology.

In one embodiment, describe the atomic Decomposition grade and decomposition level of trading volume decomposition levels, it produces one group in inner consistent trading volume decomposition levels set.The atomic Decomposition grade is represented an inseparable active set, and be used to build on the atomic Decomposition grade in inner consistent other active set and trading volume decomposition levels.

Though describe atomic Decomposition and decompose level according to decomposing trading volume, in alternate embodiment, this process can be used for calculating to other can measure the atomic Decomposition and decomposition level that commerce is measured (for example, income, profit or the market share etc.).For instance, in one embodiment, like preceding text at Figure 10 to described in Figure 13, processing logic measures calculation of atomic to decompose and/or decompose level to another measurable commerce.

Mix the reason report

Trading volume mentioned above is decomposed with trading volume decomposition level and is attempted to answer following problem: " what activity has contribution to sales volume (or other can measure commerce measure)? " Be applied to one or more products, time cycle and/or place though trading volume decomposed, with single sale spirogram modeling the time, use trading volume usually and decompose.The report of another type " reason " confirms that the trading volume that causes because of the trading volume difference of two different times between the cycle contributes.Therefore, reason attempts to answer following problem: " why trading volume rises/descend? " And define how many trading volumes changes and be attributable to each in the specific activities.In one embodiment, when when different time is directed against identical product set analysis trading volume variation (for example the sales volume with every year compares) in the cycle, the reason report is particularly useful.

The analyst will want usually to know that which activity change has caused the trading volume variation.Figure 15 is the chart of an embodiment of illustration report 1500.In Figure 15, reason report 1500 comprises beginning trading volume 1502, the amount of closing the trade 1504, trading volume contribution variation 1506, the variation 1512 of basic trading volume and model error 1510.Beginning trading volume 1502 is the trading volumes from cycle start time, and the amount of closing the trade 1504 is the trading volumes from cycle concluding time.The variation that reason report 1500 will begin trading volume 1502 and the amount of closing the trade 1504 is rendered as each variation and the variation that basic trading volume is contributed of trading volume contribution that comprises in the activity.For instance, in Figure 15, the variation of basic trading volume 1512 makes beginning trading volume 1502 increase+2.98%.

In addition, in Figure 15, the variation of activity trading volume 1506 can be positive and negative and/or is zero, and in from-4% to+3.48% scope.For instance, the trading volume that is attributable to movable NumItems is changed to+and 3.48%.Model error 1510 makes the variation of trading volume contribution increase-0.4% and-1.1%.

So known in the technology, a scheme of calculating the reason report is to calculate to the trading volume of the beginning and the amount of closing the trade to decompose report, and confirms each movable difference from these reports.This scheme is called the report of difference decomposing cause in this technology.The difference decomposing cause proposes following problem: " coming the trading volume contribution of oneself's activity how to change? "

With the difference decomposition computation is (by movable) difference of decomposition and the trading volume in the decomposition in the time cycle one contribution in the time cycle two.Yet the difference decomposing cause has following shortcoming: some activity can't natural decomposition, and therefore becomes the part of the basic trading volume in the decomposition.In one embodiment, some activities do not have the nature reference value, and can't be contained in naturally in the trading volume decomposition.These a little movable instances are underlying price or distribution.Therefore, the difference decomposing cause is shown as the basis with these effects to be changed, and it is not decomposed.

Whether irrelevant with activity based on the driver that has or do not have reference value, use these movable start and end values to calculate reason.This is called the mixing reason.Mix reason and propose following problem: " how the variation of Activity Level changes trading volume? " It does not change one's plans the analyst situation about taking place under the plan of situation about taking place and change is compared, and difference is belonged to movable variation.

Calculating the first step that mixes reason is the variation of confirming the basic trading volume between two time cycles.This of basis trading volume changes the interim fluctuation by the sales volume that under situation about directly not caused by the marketing activity of company, takes place, and/or by not in response model the activity of modeling produce.Notice that even activity does not change, the trading volume contribution also can change.For instance, trading volume can increase based on the puffing advertising of customers, or the popularity of product set can increase naturally/reduce.The variation of the trading volume that this produces owing to changing (for example the basis changes) by the driver that is not associated with activity.

Figure 14 is that explanation is predicted trading volume to the difference in different time cycle and is attributable to basic trading volume and the block diagram 1400 of an embodiment of the variation of the variation of activity.In Figure 14, beginning trading volume 1402 is the trading volumes in the cycle 1 (P1), and the amount of closing the trade 1406 is the trading volumes in the cycle 2 (P2), and both all are based on, and the activity of execution is predicted in cycle 1 and 2 respectively.In Figure 140 0, the amount of closing the trade 1406 is greater than beginning trading volume 1402.Also can be after executed and the movable identical activity of in the cycle 2, carrying out, the trading volume of predetermined period 2, thus produce the trading volume that marks by 1404.Difference between the trading volume 1404 and 1402 is the variation of two basic trading volumes between the cycle.Trading volume 1406 and 1402 variation are attributable to the variation of basic trading volume 1404, and change because of movable 1410.The trading volume 1402 that trading volume 1404 expressions are proofreaied and correct to basic variation.After adjusting to the variation of basic trading volume, the remainder that trading volume changes is attributable to movable 1410 variation.

Decompose reference value and carry out one in the decomposition algorithm (for example addition, subtraction) through the driver values among the P1 is used as, calculate the original trading volume that will belong to movable variation.In addition, distribute synergy with in the mode mentioned above any one.The variation of trading volume contribution is also referred to as the trading volume variance.

In this embodiment, the driver in this model has defined value in the cycle 1.For this reason, in mixing the reason algorithm, do not have reference value yet, and therefore will have trading volume contribution through division even do not have the driver of clear and definite reference value.

Figure 16 is used to calculate the process flow diagram that mixes reason and distribute an embodiment of synergistic process 1600.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 1600 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 16, handling frame 1602 places, said process through the processing logic access begin/amount of closing the trade and activity value and this process in employed other input information (response model etc.) begin.Processing logic is being handled the calculating beginning of frame 1604 places and the amount of closing the trade.In one embodiment, calculate the beginning and the amount of closing the trade described in the processing frame 406 among processing logic such as Fig. 4.

Handling frame 1606 places, processing logic calculates the variation of basic trading volume.As indicated above, the variation of basic trading volume can be by the interim fluctuation of abiogenous sales volume and/or by the not activity of modeling generation in response model.In one embodiment, processing logic calculates a variation of predicting that trading volume (for example, finishing or begin trading volume) is calculated basic trading volume in another cycle through the executed activity (for example, beginning or end executed activity) of use one-period.Predict that the difference between the trading volume is the variation of basic trading volume for these two.

Processing logic is carried out cycle of treatment (handling frame 1608 to 1614) with each the variation of trading volume contribution in the computational activity.Handling frame 1610 places, the processing logic activity is set to one in the start and end value, and all other activities are set to other value set.For instance, if activity of processing logic is set to starting value, all other activities of processing logic are set to end value so.Use this to be provided with, processing logic calculates the trading volume contribution of the activity with different beginning/end value.In one embodiment, processing logic through obtain with difference begin/trading volume that end value is calculated calculates trading volume contribution with the difference between the corresponding beginning/amount of closing the trade.Hereinafter with reference table 1 further describes the instance of such calculating to table 10.Cycle of treatment is being handled the end of frame 1614 places.

Handling frame 1616 places, processing logic is distributed to synergy the trading volume contribution set of in the cycle of treatment of preceding text, calculating.In one embodiment, processing logic comes to each the distribution synergistic part of calculating in the activity trading volume variance based on the absolute value of original trading volume variance.In one embodiment, make V _I, V ₂..., V _nBy just be assigned with those movable original trading volume variances of the synergistic part of calculating, and make the S be the synergy of being calculated that will distribute.In one embodiment, use equality (13) to calculate the final trading volume variance of each movable i:

$V_i^{\mathrm{Final}}=V_i+\frac{\left|V_i\right|}{{\mathrm{\&Sigma;}}_{j=1}^n\left|V_j\right|}\&CenterDot;S---\left(13\right).$

Wherein

Be the final trading volume variance of movable i, V _iBe the original trading volume variance of movable i, and S is total synergy of calculating.Processing logic is being handled frame 1618 place's computation model errors.

In one embodiment, how the instance explanation processing logic that provides of hereinafter contains from the cube of some drivers of beginning and done state through establishment and confirms trading volume " reason " variation.In this example, the active set through modeling comprises marketing, trade, price and distribution.Marketing activity comprises driver TV.Trade activity comprises driver TPR_Price and TPR_ACV.The price activity comprises driver NoPromoPrice and AverageNoPromoPrice.The distribution activity comprises driver ACV.Be used to calculate trading volume (equality (14)) with drag with these activities:

Trading volume=(1.5) * (TV)+

(2000)*(AverageNoPromoPrice-TPR_Price)*TPR_ACV/ACV+(1000)*(AverageNoPromoPrice-NoPromoPrice)*(ACV-TPR_ACV)/ACV+(150)*(ACV) (14)

In addition, in this example, the subtraction form of decomposing scheme is used with absolute synergy allocative decision.In other embodiments, use other scheme.

The beginning driver values and the comparison that finishes driver values of the activity of table 1 explanation marketing, trade, price and distribution.

Table 1. beginning and end driver values.

Use and finish driver values, processing logic calculates the amount of closing the trade.As illustrated in the table 2, the amount of closing the trade is the trading volume of 8,600 units.

Table 2. calculates the amount of closing the trade.

Processing logic switches to starting value and recomputates trading volume with this configuration from end value through making each activity, calculates in the original trading volume contribution each.For instance, processing logic is set to 100 through the TV driver and switches the market value.The gained trading volume is 8660, or-60 variation (table 3).

Table 3. calculates the trading volume of the marketing activity with beginning driver values.

For the original trading volume contribution of trade activity, processing logic makes marketing activity return to the end driver values, and trade activity is set to starting value.In this example, processing logic is set to 2.1 and 20 respectively with the driver TPR_Price and the TPR_ACV of trade.The trading volume that is calculated is 9,130, and it is-530 variation (table 4).

Table 4. calculates the trading volume of the trade activity with beginning driver values.

For the price activity, processing logic makes trade activity return to the end driver values, and the price activity is set to begin driver values.In this example, processing logic is set to 3.90 and 4.10 respectively with price driven device NoPromoPrice and AverageNoPromoPrice.The trading volume that is calculated is 8,810, and it is-210 variation (table 5).

Table 5. calculates the trading volume of the price activity with beginning driver values.

For the distribution activity, processing logic makes the price activity return to the end driver values, and the distribution activity is set to begin driver values.In this example, processing logic price driven device ACV is set to 40.The trading volume that is calculated is 7,378, and it is 1223 variation (table 6).

Table 6. calculates the trading volume of the distribution activity with beginning driver values.

In addition, processing logic uses the beginning driver values to calculate the beginning trading volume, its produce 8,250 trading volume and with 350 (tables 7) that differ from of final trading volume.

Table 7. beginning trading volume and the beginning trading volume/amount of closing the trade are poor.

Yet the summation of initial transaction amount contribution is 423 (tables 8).There is the synergy of 73 units that distribute to the contribution of individual transaction amount in this indication.

The contribution of table 8. individual transaction amount, unallocated synergy.

Through using like preceding text, list the contribution of final trading volume in the table 9 in the absolute synergy allocative decision described in Fig. 6.For instance, for marketing activity, the increment trading volume is-60, and absolute value is 60, and the summation of the absolute value that each is movable is 2023, and total synergy is for-73.Distribute through using equality (7) to calculate to the synergy of marketing activity, total synergistic institute distribution portion is-2.2.The final increment trading volume of marketing activity is-62.2.

Synergistic final trading volume contribution is distributed and comprised to table 9. synergy.

List net result in the table 10.

Table 10. net result.

As indicated above, in one embodiment, calculate to mix the reason report, it confirms not have the trading volume variance between two different trading volumes of an active set of reference value.In addition, in other embodiments, will mix reason and be applied to other active set (having activity, distribution activity of reference value etc.).

In another embodiment, calculate a plurality of grades of trading volume variance based on defined atom action set, the decomposition level that comprises activity tree and mixing reason.In this embodiment, processing logic such as preceding text are described in Figure 16 and use said defined atom action to gather to come calculation of atomic trading volume variance grade.Through using this atom trading volume variance grade, based on the higher level in the incompatible calculating trading volume of other active set variance grade, said other active set is based on said defined atom action set described in Figure 13 for processing logic such as Figure 11.In this embodiment, sue for peace to the trading volume variance rather than to trading volume contribution.

Though describe the mixing reason according to calculating the trading volume variance, in alternate embodiment, this process can be used for can measuring commerce to other and measures (for example, income, profit or the market share etc.) and calculate Calculation variance.For instance, in one embodiment, processing logic such as preceding text can be measured commerce to another and measure and calculate the trading volume variance described in Figure 16.

Compound reason report

As indicated above, use Difference Solution or hybrid plan to calculate the reason report.The report of Difference Solution reason is calculated in the activity that only has a suitable reference value to its driver, and mixes the activity that the reason report is used to have or do not have suitable reference value.In some cases, the decipher of user's preferences " variation of movable trading volume contribution " that the difference by decomposition method is provided is better than the decipher of mixed method " trading volume owing to the variation of activity changes ".In order to keep the ability of the trading volume variance report that provides on these active sets, in one embodiment, these two schemes are made up to be used to calculate the reason report.In this embodiment, Difference Solution is applied to have the activity of the driver of tool nature reference value,, and calculates the trading volume that produces by the distribution activity through subtraction and change the non-distribution activity of mixing application in the driver that does not have tool nature reference value.Figure 17 is the process flow diagram of an embodiment that is used to calculate the process 1700 of compound reason.Said process can be carried out by processing logic, and said processing logic can comprise hardware (for example, circuit, special logic, FPGA, microcode etc.), software (for example on general-purpose computing system or custom-built machine, moving), or the two combination.In one embodiment, process 700 is carried out by the data handling system 1800 of Figure 18.

Referring to Figure 17, handling frame 1702 places, said process is confirmed and will be begun through the active set that Difference Solution, mixing or other scheme are calculated through processing logic.In one embodiment, processing logic use Difference Solution is calculated the trading volume contribution of the activity of the driver with tool nature reference value.In addition, processing logic uses hybrid plan to calculate the trading volume contribution of all other non-distribution activities.Will through from begin/amount of closing the trade difference deducts Difference Solution and mixing resultant calculates the distribution activity.

For instance, in one embodiment, consider the response model of forming by following activity: TV, trade, price, underlying price and distribution.For this active set, TV and trade have reference value, and price, underlying price and distribution do not have reference value.In the compound scheme of calculating the reason report, will use Difference Solution to calculate TV and trade, will use and mix to come accounting price and underlying price, and will calculate distribution (table 11) through subtraction.

" bucket "	Calculation stages
		TV	DD
Trade	DD
		Price	H
The basis	H
		Distribute	Subtraction

Table 11. is to each the numerical procedure in the activity.

During in calculation stages each, will be during the said stage invalid driver be regarded as the reason report it does not paid close attention to.Therefore, the effect of those drivers will be the part of basic trading volume, as described in (table 12).Total basic trading volume is the set like the not appointment activity of being stated in the decomposition levels.

Effectively the basis is the active set of during each stage of calculating, handling with class basis mode.During the calculation stages not the activity in effective basis will switch its driver values.In stage, the basic merger of being stated is basic for effectively in different decomposition.In this embodiment, in calculating the process of corresponding decomposition, do not switch as the driver of the part on effective basis.In mix stages, the basis of being stated with catabolic phase in the basic different mode of being stated treated.This means that mixing the driver in the basis of stating switches between two different values.

The bucket of being stated	Employed bucket among the DiffD	Employed bucket in the mixing
			TV	?TV DD	Effectively basic H
Trade	Trade DD	Effectively basic H
			Price	Effectively basic DD	Price H
The basis	Effectively basic DD	The basis H
			Distribute	Effectively basic DD	Effectively basic H

Effective basis of each in table 12. activity.

Handling frame 1704 places, processing logic is to having the activity application Difference Solution of nature reference value.In one embodiment, processing logic decomposes to each the calculating trading volume in beginning and the end scene.Through using this two trading volumes to decompose, processing logic comes each the trading volume variance in the computational activity through the difference that obtains two trading volumes and decompose.In addition, processing logic can distribute any synergistic part of calculating in the trading volume variance each.

In the instance of the activity that in table 11, provides, processing logic will use decomposition levels (TV _DD, trade _DDWith effective basis _DD) the Difference Solution decipher, and use Difference Solution to carry out calculating.Note, for this calculating, effectively basic _DDIn activity (basis of price and distribution) do not switch.Processing logic uses equality (15) to calculate synergy:

$(V_{{\mathrm{TV}}_{\mathrm{DD}}}+V_{{\mathrm{Trade}}_{\mathrm{DD}}})-V_{{\mathrm{TV}}_{\mathrm{DD}}-\mathrm{and}-{\mathrm{Trade}}_{\mathrm{DD}}}---\left(15\right)$

Wherein

is under the situation of reference value and the trading volume of calculating at TV;

is under the situation of reference value and the trading volume of calculating in trade, and

is under the situation of reference value and the trading volume of calculating in TV and trade.Processing logic can distribute synergy in this step or in the step of back.

Handling frame 1706 places, processing logic is not to having the nature reference value or being not the activity application hybrid plan of distribution activity.In one embodiment, described in processing logic such as Figure 16 and use hybrid plan to calculate the trading volume variance.As being applied to the activity in the table 11, processing logic will use hybrid plan to price and underlying price, and carry out hybrid.In this embodiment, the activity (TV and trade) among effective basic H is not switched.Processing logic uses equality (16) to calculate synergy:

$(V_{{\mathrm{Price}}_H}+V_{{\mathrm{Base}}_H})-V_{{\mathrm{Price}}_H-\mathrm{and}-{\mathrm{Base}}_H}---\left(16\right)$

Wherein

is under the situation of one in beginning/done state and the trading volume of calculating in price;

is under the situation of one in beginning/done state and the trading volume of calculating in underlying price, and is under the situation of one in beginning/done state and the trading volume of calculating in price and underlying price.Processing logic can distribute synergy in this step or in the step of back.

Handling frame 1708 places, processing logic is as calculating the trading volume variance of distribution activity through deduct other trading volume variance of during handling frame 1704 and 1706, calculating from total prediction trading volume in equality (17):

${\mathrm{VV}}_{\mathrm{Dist}}=V_{\mathrm{Predicted}}-\mathrm{\&Sigma;}{\mathrm{VV}}_i-{\mathrm{Synergy}}_{\mathrm{DD}}-{\mathrm{Synergy}}_H---\left(17\right)$

VV wherein _DistBe distribution trading volume variance, V _PredictedBe the prediction trading volume, ∑ VV _iBe to use the summation of the trading volume variance of Difference Solution and hybrid plan calculating, synergy _DDThe synergy that is to use Difference Solution to calculate, and synergy _HThe synergy that is to use hybrid plan to calculate.

In alternate embodiment, processing logic does not use the subtraction stage to calculate compound reason.In this embodiment, processing logic uses the difference of decomposition method to calculate the trading volume contribution according to those activities of assigning poor (DD) that give decomposition group.

" bucket "	Calculation stages
		TV	DD
Trade	DD
		Price	H
Distribute	H

Table 13. do not have subtraction to each the numerical procedure in the activity in the compound reason.

In two time cycles, the driver that is associated with activity among the DD is set to its decomposition value.The result predicts the new model of the trading volume with the movable contribution of DD.With respect to unallocated those activities of giving DD will mix the reason algorithm application in this new model provide the movable trading volume of non-DD contribute and the variation of basic trading volume both.

In more detail, calculating does not have the compound reason of subtraction in following three steps:

1. calculate the decomposition of two time cycles with respect to the activity among the DD.The difference that obtains between the corresponding original trading volume contribution provides these movable original trading volume contributions.

2. be set to its reference value corresponding to the All Drives of the activity among the DD new model can be provided.Calculate the mixing reason based on this model.For instance, described in Figure 16, calculate the mixing reason like preceding text.This reason provides the original trading volume contribution of the activity that mixes in the reason and the variation of basic trading volume.

3. distribute synergy (poor between the summation of all original trading volumes contributions and the difference of predicting trading volume through one in the described mechanism of preceding text.

In another embodiment, gather, comprise the decomposition level of activity tree and a plurality of grades that compound reason is calculated the trading volume variance based on defined atom action.In this embodiment, processing logic such as preceding text are described in Figure 17 and use said defined atom action to gather to come calculation of atomic trading volume variance grade.Through using this atom trading volume variance grade, processing logic is based on the higher level in the incompatible calculating trading volume of other active set variance grade, and said other active set is based on like Figure 11 to be gathered to the said defined atom action described in Figure 13.In this embodiment, sue for peace to the trading volume variance rather than to trading volume contribution.

Though describe compound reason according to calculating the trading volume variance, in alternate embodiment, this process can be used for calculating to other measurable commerce to be measured the atomic Decomposition of (for example, income, profit or the market share etc.) and decomposes level.For instance, in one embodiment, processing logic such as preceding text at Figure 10 described in Figure 13 and calculate the atomic Decomposition of measuring to another measurable commerce and/or decompose level.

Figure 18 calculates the block diagram that trading volume is decomposed, atomic Decomposition/trading volume is decomposed the data handling unit (DHU) assembly 800 of level, mixing reason and/or compound reason.Data handling system is not limited to multi-purpose computer, multiprocessor computer, passes through some computing machines of network coupled etc.In Figure 18, system 1800 comprises trading volume decomposing module 1802, decomposition level module 1804, synergy module 1806, compound reason module 1808, trading volume module 1810 and mixes reason module 1812.Trading volume module 1810 accesses input and calculating trading volume result.In one embodiment, input comprises each the value in response model, active set and the said activity, and is as shown in Figure 2.In one embodiment, trading volume decomposing module 1802 with mix reason module 1812 guiding trading volume modules and calculate one or more in basic trading volume, expection trading volume, the trading volume contribution etc.Synergy module 1806 is calculated and is distributed synergy, described in Fig. 7.In one embodiment, trading volume decomposing module 1802 is calculated and the distribution synergy with mixing reason module 1812 guiding trading volume modules.

Trading volume decomposing module 1802 comprises basic trading volume module 1820, original trading volume contribution module 1822, expection trading volume module 1824, final trading volume contribution module 1826 and load module 1828.Basis trading volume module 1820 is as described in the processing frame 406 of Fig. 1 and calculate basic trading volume.Original trading volume contribution module 1822 is as described in the processing frame 410 to 418 of Fig. 4 and calculate each the original trading volume contribution in the activity with reference value.Described in expection trading volume module 1824 as the processing frame 406 of Fig. 4 and calculate and expect trading volume.In one embodiment, basic trading volume module 1820, original trading volume contribution module 1822 and expection trading volume module booting trading volume module 1810 are calculated suitable trading volume.Final trading volume contribution 1826 increases definite synergy of being distributed, and adds it in the original trading volume contribution each, described in the processing frame 420 of Fig. 4.In one embodiment, final trading volume module 1826 uses synergy module 1806 definite synergies to distribute.Load module access input is described in the processing frame 402 of Fig. 4.

Decompose level module 1804 and comprise atomic Decomposition module 1830 and higher level decomposing module 1832.Atomic Decomposition module 1830 definition and calculation of atomic decomposition levels are described in Figure 10.Higher level decomposing module 1832 is calculated the grade that trading volume is decomposed based on atomic Decomposition, like Figure 11 described in Figure 13.

Synergy module 1806 comprises total synergy module 1840 and synergy contribution module 1842.Total synergy module 1840 is calculated total synergy based on increment trading volume and original trading volume contribution, described in the processing frame 604 of Fig. 6.Indivedual synergies contributions of each in the 1842 definite input activities of synergy contribution module are described in the processing frame 606 of Fig. 6.In one embodiment, synergy contribution module 1842 is confirmed indivedual synergy contributions based on the absolute value of each original trading volume contribution.

Compound reason module 1810 comprises mixing module 1850, Difference Solution module 1852 and distribution reason module 1854.Mixing module calculates the trading volume variance of the non-distribution activity that does not have reference value, described in the processing frame 1706 of Figure 17.Difference Solution module 1852 is calculated the trading volume variance of the non-distribution activity with reference value, described in the processing frame 1704 of Figure 17.Distribution reason module 1854 is calculated the trading volume variance of distribution activity, described in the processing frame 1708 of Figure 17.

Mixing reason module 1812 comprises and beginning/amount of closing the trade module 1860, basic trading volume module variations module 1862, original trading volume contribution module 1864, synergy module 1866, model error module 1868 and load module 1870.Begin/amount of the closing the trade module calculating beginning and the amount of closing the trade, described in the processing frame 1604 of Figure 16.The basis changes trading volume module 1862 and calculates basic trading volume variation, described in the processing frame 1606 of Figure 16.Original trading volume contribution module calculates original trading volume contribution to be changed, described in the processing frame 1608 to 1614 of Figure 16.Synergy module 1866 is to each the calculating synergy in the activity, described in the processing frame 1616 of Figure 16.Model error module 1868 computation model errors are described in the processing frame 1618 of Figure 16.Load module 1870 access input parameters are described in the processing frame 1602 of Figure 16.

The described method of preceding text is calculated the reason report to the right prediction trading volume (scene) of single coupling (for example to the single product in the single place in two different weeks).Be directed against the reason report of the right scene set of coupling (for example, two different all interior a plurality of product/places) if desired, calculate the reason report of all individual transaction amounts so, and add the trading volume contribution of respective activity.

Process described herein can constitute one or more programs of being made up of machine-executable instruction.Describe said process with reference to figure 4, Fig. 6, Figure 10 to the process flow diagram among Figure 13, Figure 16 and Figure 17 and make the those skilled in the art can develop this a little programs, comprise in order to go up the said instruction of carrying out the operation of showing by the logical process frame table (action) at the adaptive machine of putting of ECDC (processor of said machine is carried out from machine-readable medium (for example RAM (for example DRAM), ROM, non-volatile storage medium (for example hard disk drive or CD-ROM) etc.) instruction).Machine-executable instruction available computers programming language is write, and maybe can be contained in the firmware logic or in the hardware circuit.If the programming language to meet Recognized Standards is write, on the multiple hardwares platform and to the interface that arrives several operation systems, carry out said instruction so.In addition, do not describe the present invention, will understand, can use multiple programming language to implement like teaching of the present invention described herein with reference to any certain programmed language.Software (for example, program, process, process, application program, module, the logic that will be in addition, a kind of form or another form ...) say to become to take to move or cause that the result is common in this technology.These a little expression only are the simple modes that the statement machine is carried out action or born results the processor of the execution results in machines of software.To further understand, can without departing from the scope of the invention more or less process be incorporated in the process illustrated in the process flow diagram, and the layout of the frame that this paper showed and described not hint certain order.

Figure 19 shows the some computer systems 1900 that are coupled through network 1902 (for example the Internet).As used herein term " the Internet " refers to and uses some agreement (network of network of ICP/IP protocol and possible other agreement HTTP (HTTP) of the HTML(Hypertext Markup Language) document that constitutes WWW (web) (for example to) for example.The physical connection of the Internet and Internet protocols and signal procedure is that the those skilled in the art is well-known.Access to the Internet 1902 is provided by ISP (ISP) (for example ISP1904 and 1906) usually.User on the FTP client FTP (for example, client computer system 1912,1916,1924 and 1925) obtains the access to the Internet through ISP (for example ISP 1904 and 1906).Access to the Internet allows user's exchange message, reception and the send Email of client computer system and checks document, the document of for example having prepared with html format.These documents usually by be regarded as the Internet " on " web server (for example the web server 1908) provide.Usually, these web servers are provided by ISP (for example ISP 1904), but so well-known in the technology, under computer system neither the situation of ISP, said system can be set and be connected to the Internet.

Web server 1908 is generally at least one computer system of operating as server computer system, and operates with the agreement of WWW through being configured to, and is coupled to the Internet.Optional is, web server 1908 can be for FTP client FTP the part to the ISP of the access of the Internet to be provided.

Web server 1908 is shown as is coupled to server computer system 1910, itself be coupled to web content 1912, web content 1912 can be regarded as a kind of media database of form.To understand; Though show two computer systems 1908 and 1910 among Figure 19; But web server system 1908 can be a computer system with different software assembly, the server functionality that said component software provides the web server functionality and provided by the server computer system that will further describe hereinafter 1910 with server computer system 1910.

Client computer system 1912,1916,1924 and 1926 can be checked the HTML page or leaf that is provided by web server 1908 through suitable web browsing software separately.ISP 1904 provides the Internet connectivity that arrives client computer system 1912 through modem interface 1914 (it can be regarded as the part of client computer system 1912).Client computer system can be personal computer system, network computer, Web TV system, handheld apparatus or other this type of computer system.Similarly, ISP 1906 provides the Internet connectivity for FTP client FTP 1916,1924 and 1926, but as shown in Figure 19, said connection is different for these three computer systems.Client computer system 1916 is coupled through modem interface 1918, and client computer system 1924 and 1926 is parts of LAN.Though Figure 19 is shown as common modulator-demodular unit with interface 1914 and 1918 "; will understand, each in these interfaces can be analog modem, isdn modem, cable modem, satellites transmits interface or is used for a coupled computer systems other interface to other computer system.Client computer system 1924 and 1916 is coupled to LAN 1922 through network interface 1930 and 1932, and network interface 1930 and 1932 can be Ethernet or other network interface.LAN 1922 also is coupled to gateway computer system 1920, and it can be LAN fire wall and other internet related services are provided.This gateway computer system 1920 is coupled to ISP 1906, so that the Internet connectivity that arrives client computer system 1924 and 1926 to be provided.Gateway computer system 1920 can be the General Server computer system.And web server system 1908 can be conventional server computer system.

Perhaps, as as everyone knows, server computer system 1928 can be directly coupled to LAN1922 through network interface 1934, so that file 1936 and other service are offered client 1924,1926, and need not to be connected to the Internet through gateway system 1920.In addition, any combination of FTP client FTP 1912,1916,1924,1926 can be used LAN 1922, the Internet 1902 or as the combination of communication medium and be connected in the peer-to-peer network together.Usually, peer-to-peer network on the network of a plurality of machines, for storage and retrieval, and need not to use central server with DATA DISTRIBUTION.Therefore, each peer network node can be incorporated into has both functions of the described client and server of preceding text.

Following description to Figure 20 is intended to the general introduction of carrying out computer hardware with other operating assembly of the described process of the present invention of preceding text to suitable is provided, but is not intended to the environment of restricted application.The those skilled in the art will understand at once, and available other computer system configurations (comprise STB, handheld apparatus, multicomputer system, based on consumer electronics microprocessor or programmable, network PC, microcomputer, mainframe computer etc.) is put into practice embodiments of the invention.Embodiments of the invention also can be put into practice in DCE, in DCE, execute the task through the teleprocessing device that links via communication network (for example peer-to-peer network foundation structure).

Figure 20 is can be at an instance of the conventional computer system of using in aspect one or more of the present invention.Computer system 2000 is situated between via modulator-demodular unit or network interface 2002 and receives external system.To understand, can modulator-demodular unit or network interface 2002 be regarded as the part of computer system 2000.This interface 2002 can be analog modem, isdn modem, cable modem, token ring interface, satellites transmits interface or is used for a coupled computer systems other interface to other computer system.Computer system 2002 comprises processing unit 2004, and it can be conventional microprocessor, for example intel pentium microprocessor or the Power PC of Motorola microprocessor.Storer 2008 is coupled to processor 2004 through bus 2006.Storer 2008 can be dynamic RAM (DRAM), and also can comprise static RAM (SRAM) (SRAM).Bus 2006 is coupled to storer 2008 with processor 2004, and also is coupled to Nonvolatile memory devices 2014, and is coupled to display controller 2010, and is coupled to I/O (I/O) controller 2016.Display controller 2010 is controlled the demonstration on the display device 2012 in a usual manner, and display device 2012 can be cathode ray tube (CRT) or LCD (LCD).Input/output device 2018 can comprise keyboard, disc driver, printer, scanner and other input and output device, comprises mouse or other indicator device.Display controller 2010 can be used conventional well-known technology implementation with I/O controller 2016.Digital image input device 2020 can be digital camera, and it is coupled to I/O controller 2016, will be input in the computer system 2000 from the image of digital camera so that allow.Nonvolatile memory devices 2014 is generally the memory storage that is used for mass data of magnetic hard-disk, CD or another form.In computer system 2000, during the executive software, be written in the storer 2008 through direct memory access (DMA) process some with these data.The those skilled in the art will recognize at once; What term " computer-readable media " and " machine-readable medium " comprised any kind can be by processor 2004 or by the memory storage of other data handling system (for example cellular phone or personal digital assistant or MP3 player etc.) access, and also contains the carrier wave that data-signal is encoded.

Network computer is the computer system that can use with embodiments of the invention of another type.Network calculations does not comprise hard disk or other mass storage device usually, and executable program is to connect to be loaded into the storer 2008 from network to carry out for processor 2004.Web TV system known in this technology also is regarded as computer system according to an embodiment of the invention, but it possibly lack in the characteristic shown in Figure 20 some, and for example some inputs or outputs device.The bus that typical computer will comprise at least one processor, storer usually and storer will be coupled to processor.

To understand, computer system 2000 is instances with many possible computer system of different frameworks.For instance; Personal computer based on the Intel microprocessor has a plurality of buses usually; One of them can be I/O (I/O) bus that is used for peripherals, and direct connection processing device 2004 of bus and storer 2008 (being commonly referred to memory bus).Bus links together via axle assemble, and axle assemble is carried out translating of any necessity because of different bus protocols.

Also will understand, computer system 2000 is by operating system software control, operating system software include file management system, and disc operating system (DOS) for example, it is the part of operating system software.An instance of the file management system software that operating system software is associated with it is the operating system family from the Microsoft in Redmond, State of Washington city that is called Windows (WINDOWS OPERATING SYSTEM), with and the file management system that is associated.Usually file management system is stored in the Nonvolatile memory devices 2014, and cause processor 2004 executive operating system input and output data and data storage (comprised with file storage on Nonvolatile memory devices 2014) required exercises in storer.

Alternate embodiment

Computer-implemented method comprises many activities of access response model and first; Said response model and more than first activity are used to calculate commerce and measure decomposition; Wherein more than first activity defines the commercial atomic Decomposition grade of measuring; And wherein the atomic Decomposition grade is to use the base level in the set of different decomposition of different a plurality of activities, and the set of said different decomposition is consistent with the atomic Decomposition grade.

Computer-implemented method further comprises uses response model and more than first activity to calculate basic commercial measurement value, and wherein each in more than first activity is set to the corresponding reference value.

Computer-implemented method further comprises uses response model to be set to the state of value opposite with said activity through other activity that activity is set in one in corresponding reference and the currency and more than first activity, and each that calculate in more than first activity first is contributed what commerce was measured.

Computer-implemented method further comprises uses more than second activity to calculate second decomposition levels that commerce is measured, and wherein more than second activity is based on more than first activity, and second decomposition is based on first decomposition.

Computer-implemented method further comprises uses more than the 3rd activity to calculate the 3rd decomposition levels that commerce is measured; Wherein more than the 3rd activity is based on more than the 4th activity; And the 3rd decomposition is based on lower grade decomposes, and wherein more than the 4th activity decomposed corresponding to lower grade.

Computer-implemented method, wherein each in more than first activity has reference value.

Computer-implemented method, wherein the reference value of each in more than first activity increases by zero contribution to the atomic Decomposition grade of said activity.

Computer-implemented method, wherein each in more than first activity is inseparable movable how much.

Computer-implemented method, wherein commercial measuring is one in sales volume, income, profit and the market share.

Though described various embodiment of the present invention, alternate embodiment of the present invention can be operated in a different manner.For instance; Though the process flow diagram among the figure is showed the certain order of the operation that some embodiment of the present invention is performed; But should be understood that this order is exemplary (for example, alternate embodiment can different order executable operations, make up some operation, make some operation overlap etc.).

Though described the present invention according to some embodiment, those skilled in the art will realize that to the invention is not restricted to described embodiment, but the modification in the spirit that can be used on appended claims and the scope and change and put into practice.Therefore, describing content will be regarded as illustrative rather than restrictive.

Claims (21)

1. computer-implemented method, it comprises:

Access response model and a plurality of activity, said response model and said a plurality of activity are used for calculating expection commerce to be measured, and each in wherein said a plurality of activities has reference value and executed value; And

Use said response model to be set to other activity in one in corresponding reference value and the executed value and the said a plurality of activity based on first activity and be set to calculate in said a plurality of activity each to the contribution that said commerce is measured with the said first movable opposite state of value, each in the said contribution of wherein said calculating and response model type have nothing to do.

2. it is one in sales volume, income, profit and the market share that computer-implemented method according to claim 1, wherein said commerce are measured.

3. computer-implemented method according to claim 1, wherein said first activity is set to said reference value, and said other activity in said a plurality of activity is set to said executed value.

4. computer-implemented method according to claim 1, wherein said first activity is set to said executed value, and said other activity in said a plurality of activity is set to said reference value.

5. computer-implemented method according to claim 1 is wherein calculated said contribution and further is included in said first activity and is in and calculates the prediction sales volume under the situation of one in said corresponding reference value and the currency.

6. computer-implemented method according to claim 5 is wherein calculated said contribution and is comprised that further calculating equals the original trading volume contribution that the executed sales volume deducts said prediction sales volume.

7. computer-implemented method according to claim 1, it further comprises:

Have under the situation of said reference value in said a plurality of activities, use said response model to calculate basic commerce and measure.

8. computer-implemented method according to claim 1, it further comprises:

Distribute the synergistic part calculated in the said contribution each.

9. computer-implemented method according to claim 8, wherein said distribution is based on the absolute value of said contribution.

10. computer-implemented method according to claim 8, the said synergistic said part that calculates of wherein said distribution is based on following formula:

$V_i^{\mathrm{Final}}=V_i+\frac{\left|V_i\right|}{{\mathrm{\&Sigma;}}_{j=1}^n\left|V_j\right|}\&CenterDot;S$

Wherein

Be the final contribution of movable i, V _iBe the original contribution of movable i, and S is the said synergy that calculates.

11. a machine-readable storage media, it has executable instruction and comprises following method of operating with the cause processor execution:

Access response model and a plurality of activity, said response model and said a plurality of activity are used for calculating expection commerce to be measured, and each in wherein said a plurality of activities has reference value and executed value; And

Use said response model to be set to other activity in one in corresponding reference value and the executed value and the said a plurality of activity based on first activity and be set to calculate in said a plurality of activity each to the contribution that said commerce is measured with the said first movable opposite state of value, each in the said contribution of wherein said calculating and response model type have nothing to do.

12. it is one in sales volume, income, profit and the market share that machine-readable storage media according to claim 11, wherein said commerce are measured.

13. machine-readable storage media according to claim 11, it further comprises:

Distribute the synergistic part calculated to be based on the absolute value of said contribution in the said contribution each.

14. machine-readable storage media according to claim 11 is wherein calculated said contribution and further is included in said first activity and is in and calculates the prediction sales volume under the situation of one in said corresponding reference value and the currency.

15. an equipment, it comprises:

Load module, it is in order to access response model and a plurality of activity, and said response model and said a plurality of activity are used for calculating expection commerce to be measured, and each in wherein said a plurality of activities has reference value and executed value; And

Original contribution module; It is set to calculate in said a plurality of activity each to the contribution that said commerce is measured with the said first movable opposite state of value in order to use said response model to be set to other activity in one in corresponding reference value and the executed value and the said a plurality of activity based on first activity, and each in the said contribution of wherein said calculating and response model type have nothing to do.

16. it is one in sales volume, income, profit and the market share that equipment according to claim 15, wherein said commerce are measured.

17. equipment according to claim 15, it further comprises:

Final contribution module, it is in order to distribute a synergistic part that is calculated based on the absolute value of said contribution in the said contribution each.

18. a system, it comprises:

Processor;

Storer, it is coupled to said processor via bus; And

From carrying out to cause said processor to carry out the process of following operation of said storer by said processor; Access response model and a plurality of activity; Said response model and said a plurality of activity are used for calculating expection commerce to be measured, and each in wherein said a plurality of activities has reference value and executed value; And

Use said response model to be set to the state of value movable opposite with said first based on other activity that first activity is set in one in corresponding reference value and the executed value and the said a plurality of activity; Calculate in said a plurality of activity each to the contribution that said commerce is measured, each in the said contribution of wherein said calculating and response model type are irrelevant.

19. it is one in sales volume, income, profit and the market share that system according to claim 18, wherein said commerce measure.

20. the synergistic part that system according to claim 18, wherein said process further cause said processor to be calculated to each distribution in the said contribution is based on the absolute value of said contribution.

21. a computer-implemented method, it comprises:

Access response model and more than first activities, said response model and more than first activity are used to calculate commerce and measure decomposition, wherein said more than first atomic Decomposition grades that the said commerce of movable definition is measured; And wherein said atomic Decomposition grade is to use the base level of the different decomposition set of different a plurality of activities, and the set of said different decomposition is consistent with said atomic Decomposition grade.

Images