CN106875027A

CN106875027A - The Forecasting Methodology and device of resource request value, the Forecasting Methodology of trading volume

Info

Publication number: CN106875027A
Application number: CN201610393294.8A
Authority: CN
Inventors: 喻继银; 潘晓峰
Original assignee: Alibaba Group Holding Ltd
Current assignee: Advanced New Technologies Co Ltd; Advantageous New Technologies Co Ltd
Priority date: 2016-06-06
Filing date: 2016-06-06
Publication date: 2017-06-20
Anticipated expiration: 2036-06-06
Also published as: CN106875027B

Abstract

This application discloses the Forecasting Methodology and device of a kind of Internet resources value request, the accuracy for improving prediction Internet resources value request, the method includes：Historical requests data are obtained, and time series data is determined according to the historical requests data, included in the time series data：The value request of target resource in some unit intervals；According to the time series data, time series trend information is determined, and the initial predicted value that target resource is asked is determined according to the time series trend information；Obtain the history exchange rate data of the target resource and homegrown resource；According to the history exchange rate data, exchange rate tendency information is determined, and determine adjustment factor according to preset rules；According to the initial predicted value and the adjustment factor, the value request of the target resource in prediction unit interval.The application is also disclosed a kind of Forecasting Methodology of internet business amount.

Description

The Forecasting Methodology and device of resource request value, the Forecasting Methodology of trading volume

Technical field

The application is related to technical field of Internet information, more particularly to a kind of Internet resources value request Forecasting Methodology and Device, and a kind of Forecasting Methodology of internet business amount.

Background technology

With the development of Internet information technique, various Internet resources (such as network storage resource, the network bandwidth are occurred in that Resource, network computing resources etc., abbreviation resource).Based on the characteristic of " resource ", should generally avoid it is idle for a long time, either oneself Fang Liyong, or other party is utilized, and resource is in by use state, so that the efficiency of maximum resource. In internet, different homegrown resources, but not necessarily one's own needs may be possessed in many ways, it is possible to By way of exchange resource, each takes what he needs.Such as, the Netowrk tape of certain bandwidth is exchanged for the network storage resource of certain capacity Resource wide.Thus the resource management system for managing resource is occurred in that.The system can manage multiple resources, it is also possible to certainly There is Resource Exchange target resource, after the resource request of user is received, if system possesses the corresponding resource of request in itself, just Homegrown resource directly can be supplied to user, if do not had in itself, can go to exchange by homegrown resource and ask corresponding Target resource, and it is supplied to user.

Because resource category is various, the supply-demand relationship in internet for different types of resource can also be changed, so In Resource Exchange, certain exchange rate will be occurred according to supply-demand relationship etc. " market factor ", such as, and 1TB network storages money The 1 month right to use in source can exchange 1 month right to use of 100MB network bandwidth resources, but if the network left unused in internet Storage resource increases, and now network storage resource occurs in that drug on the market, it is possible to if causing to exchange 100MB Netowrk tapes again 1 month right to use of resource wide, it is necessary to 1 month right to use of 1.5TB network storage resources.In " Internet resources market ", The exchange rate of various resources may all change at any time.The unstable characteristic that resource-based exchange rate has, resource management System may in advance be swapped according to user to the demand of resource, that is, in advance by homegrown resource be exchanged for target money Source, after the resource request of user is received, it is possible to will directly ask corresponding target resource to be supplied to user.

Prior art, when being exchanged in advance resource, the resource value of exchange judges merely by manual work The situation of change of exchange rate and the situation of change of user's request value, subjectivity determine next unit interval or under several units Each target resource value request of time period, and exchange rate trend, so as to be exchanged in advance.But manual type prediction money Source exchange rate and target resource value request, it is clear that it is accurate to lose.

The content of the invention

The embodiment of the present application provides a kind of Forecasting Methodology of Internet resources value request, for improving prediction Internet resources The accuracy of value request.

The embodiment of the present application provides a kind of prediction meanss of Internet resources value request, for improving prediction Internet resources The accuracy of value request.

The embodiment of the present application provides a kind of Forecasting Methodology of internet business amount, for improving prediction internet business amount Accuracy.

The embodiment of the present application uses following technical proposals：

A kind of Forecasting Methodology of Internet resources value request, including：

Historical requests data are obtained, and time series data, the time sequence are determined according to the historical requests data Included in column data：The value request of target resource in some unit intervals；

According to the time series data, time series trend information is determined, and according to the time series trend information Determine the initial predicted value of target resource request；

Obtain the history exchange rate data of the target resource and homegrown resource；

According to the history exchange rate data, exchange rate tendency information is determined, and determine adjustment factor according to preset rules；

According to the initial predicted value and the adjustment factor, the request of the target resource in prediction unit interval Value.

Preferably, according to the time series data, time series trend information is determined, and become according to the time series Gesture information determines the initial predicted value of target resource request, including：

According to the time series data, the long-term trend information of the time series data is determined；

According to the long-term trend information, using autoregressive moving-average model, the first of the target resource request is determined Step predicted value.

Preferably, according to the time series data, the long-term trend information of the time series data is determined, including：

According to the time series data, long-term trend information, the seasonal trend letter of the time series data are determined Breath and Stochastic Trends information；Then

According to the long-term trend information, using autoregressive moving-average model, the first of the target resource request is determined Step predicted value, including：

According to the long-term trend information and Stochastic Trends information, using autoregressive moving-average model, first is determined Predicted value；

According to the seasonal trend information, using third index flatness, the second predicted value is determined；

According to first predicted value and second predicted value, the tentative prediction of the target resource request is determined Value.

Preferably, according to the history exchange rate data, determine exchange rate tendency information, and determined to adjust according to preset rules Section coefficient, including：

According to the history exchange rate data, using moving average, the first exchange rate tendency information is determined, and according to One preset rules, determine the first adjustment factor；And/or

According to the history exchange rate data, using logarithm period power law model, the second exchange rate tendency information is determined, And according to the second preset rules, determine the second adjustment factor；

According to first adjustment factor and/or second adjustment factor, adjustment factor is determined.

Preferably, according to the initial predicted value and the adjustment factor, the final of the target resource request is determined Predicted value, including：

According to the initial predicted value, the adjustment factor and special time period coefficient, institute in prediction unit interval State the value request of target resource.

Preferably, historical data is obtained, and time series data is determined according to the historical data, including：

Historical requests data are obtained, and pending time series data is determined according to the historical requests data；

Missing values treatment is carried out to the pending time series data, time series data is determined.

Preferably, methods described also includes：

According to the value request of the target resource in the unit interval for predicting, target is exchanged by homegrown resource and is provided Source；

When in the unit interval, and reach the default period of the day from 11 p.m. to 1 a.m and carve, obtain and carve the mesh by the default period of the day from 11 p.m. to 1 a.m Mark the real-time accumulated value request of resource；

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved Accumulative accounting is that in history unit interval, the history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m adds up value request right Answer the ratio of value request in unit interval；

According to the real-time progressive value request and the accumulative accounting of the predetermined time, in the residing unit interval of prediction Value request；

According to the value request of the target resource in the unit interval for predicting, during with the residing unit for predicting Between in section the target resource value request, determine switch-activity.

Preferably, the accumulative accounting of history that the default period of the day from 11 p.m. to 1 a.m is carved is in multiple history unit intervals, by described pre- If the period of the day from 11 p.m. to 1 a.m carves the average value of history accumulative value request ratio of value request in correspondence unit interval of the target resource.

Preferably, the accumulative accounting of history that the default period of the day from 11 p.m. to 1 a.m is carved is the list for having identical characteristics with residing unit interval In the time period of position, the accumulative value request of history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m please in correspondence unit interval The ratio of evaluation.

Preferably, methods described also includes：

According to historical requests data, it is determined that default next period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, what default next period of the day from 11 p.m. to 1 a.m was carved The accumulative accounting of history is that in history unit interval, the history for carving the target resource by default next period of the day from 11 p.m. to 1 a.m adds up please The ratio of evaluation value request in correspondence unit interval；

According to the value request of the target resource in the residing unit interval for predicting, and default next period of the day from 11 p.m. to 1 a.m The accumulative accounting of history is carved, the value request of the target resource is carved in prediction by default next period of the day from 11 p.m. to 1 a.m；

When default next period of the day from 11 p.m. to 1 a.m quarter is reached, the reality that the target resource is carved by default next period of the day from 11 p.m. to 1 a.m is obtained When accumulative value request；

According to the value request that the target resource is carved by default next period of the day from 11 p.m. to 1 a.m of prediction, with acquisition by described Default next period of the day from 11 p.m. to 1 a.m carves the real-time accumulated value request of the target resource, determines switch-activity.

A kind of prediction meanss of Internet resources value request, the device includes：First acquisition unit, the first determining unit, Second acquisition unit, the second determining unit and predicting unit, wherein,

The first acquisition unit, for obtaining historical requests data, and when being determined according to the historical requests data Between sequence data, included in the time series data：The value request of target resource in some unit intervals；

First determining unit, for according to the time series data, determining time series trend information, and according to The time series trend information determines the initial predicted value of target resource request；

The second acquisition unit, the history exchange rate data for obtaining the target resource and homegrown resource；

Second determining unit, for according to the history exchange rate data, determining exchange rate tendency information, and according to Preset rules determine adjustment factor；

The predicting unit, for according to the initial predicted value and the adjustment factor, predicting in unit interval The value request of the target resource.

Preferably, first determining unit, specifically for：

According to the time series data, long-term trend information, the seasonal trend letter of the time series data are determined Breath and Stochastic Trends information；

Preferably, second determining unit, specifically for：

Preferably, the predicting unit, specifically for：

Preferably, the first acquisition unit, specifically for：

Preferably, described device also includes：Monitoring unit, specifically for：

Preferably, the monitoring unit, specifically for：

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved Accumulative accounting is in multiple history unit intervals, by the accumulative value request of history that the default period of the day from 11 p.m. to 1 a.m carves the target resource The average value of the ratio of value request in correspondence unit interval.

Preferably, the monitoring unit, specifically for：

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved Accumulative accounting is have in the unit interval of identical characteristics with residing unit interval, and the mesh is carved by the default period of the day from 11 p.m. to 1 a.m Mark the ratio of history accumulative value request value request in correspondence unit interval of resource.

Preferably, the monitoring unit, is additionally operable to：

A kind of Forecasting Methodology of internet business amount, including：

Historical trading data is obtained, and time series data, the time sequence are determined according to the historical trading data Included in column data：Target currency trading volume in some unit intervals；

According to the time series data, time series trend information is determined, and according to the time series trend information Determine the initial predicted value of target currency trading volume；

Obtain historical rate data；

According to the historical rate data, exchange rate tendency information is determined, and determine adjustment factor according to preset rules；

According to the initial predicted value and the adjustment factor, the transaction of the target currency in prediction unit interval Amount.

Above-mentioned at least one technical scheme that the embodiment of the present application is used can reach following beneficial effect：Gone through by basis The time series data that history request data is determined, determines time series trend information, and determine the first of target resource request Step predicted value, and by history exchange rate data, adjustment factor is determined, it is right further according to initial predicted value and adjustment factor The value request of target resource is predicted, and solves the problems, such as that the artificial subjective judgement of prior art causes accuracy relatively low, improves The accuracy of prediction target resource value request.In addition, the value request and prediction that pass through real-time estimate and monitoring residing time period Value between discrepancy, determine switch-activity, and by real-time estimate and monitor the value request and reality that are accumulated to predetermined time Border adds up value request, determines switch-activity, and homegrown resource and target resource that can be in time to possessing be adjusted, and improves money The utilization rate in source.

Brief description of the drawings

Accompanying drawing described herein is used for providing further understanding of the present application, constitutes the part of the application, this Shen Schematic description and description please does not constitute the improper restriction to the application for explaining the application.In the accompanying drawings：

Fig. 1 is the schematic flow sheet of the Forecasting Methodology of the Internet resources value request that the embodiment of the present application 1 is provided；

Fig. 2 is the schematic diagram of the Forecasting Methodology of the Internet resources value request that the embodiment of the present application 1 is provided；

Fig. 3 is the monitor in real time that the embodiment of the present application 1 is provided and determines the schematic flow sheet of the method for switch-activity；

Fig. 4 is the schematic diagram of the target resource value request real-time estimate that the embodiment of the present application 1 is provided；

Fig. 5 is the structured flowchart of the prediction meanss of the Internet resources value request that the embodiment of the present application 2 is provided；

Fig. 6 is the schematic flow sheet of the Forecasting Methodology of the internet business amount that the embodiment of the present application 3 is provided；

Fig. 7 is the monitor in real time that the embodiment of the present application 3 is provided and determines the schematic flow sheet of the method for exchange behavior.

Specific embodiment

To make the purpose, technical scheme and advantage of the application clearer, below in conjunction with the application specific embodiment and Corresponding accompanying drawing is clearly and completely described to technical scheme.Obviously, described embodiment is only the application one Section Example, rather than whole embodiments.Based on the embodiment in the application, those of ordinary skill in the art are not doing Go out the every other embodiment obtained under the premise of creative work, belong to the scope of the application protection.

Below in conjunction with accompanying drawing, the technical scheme that each embodiment of the application is provided is described in detail.

Embodiment 1

As it was previously stated, prior art is based on artificial subjective judgement prediction Resource Exchange rate and target resource request Value, such as, have been found that 1 month right to use for wanting to exchange 100MB network bandwidth resources, with 1 month of increasing capacity The network storage resource of the right to use goes to exchange, and the value request of network bandwidth resources is more and more, so, it is possible to hand in advance Switching network bandwidth resources, meet demand of the user to network bandwidth resources, but the obvious accuracy of manual type is relatively low.So being based on This defect, current inventor provides a kind of Forecasting Methodology of Internet resources value request, please for improving prediction Internet resources The accuracy of evaluation, the idiographic flow schematic diagram of the method comprises the steps as depicted in figs. 1 and 2：

Step 11：Historical requests data are obtained, and time series data is determined according to historical requests data.

When Internet resources value request is predicted, it is desirable to have based on historical data, so this step can first go to obtain Historical requests data are taken, such as, in passing 1 year, the data of all request Internet resources all there may be all the time Comprising the time, ID, the value request of target resource record, for the ease of being subsequently predicted, can be by historical requests Data are pre-processed, and obtain time series data, and time series refers to the time order and function that the numerical value of same type occurs by it The ordered series of numbers of order arrangement.So, in this step, different target resources can be divided into different time series numbers According to.Such as 1 month right to use of 100MB network bandwidth resources, there can be a time series data, wherein can wrap Contain：The value request of target resource in some unit intervals, such as, unit interval is set as day, then time series data Can be just every day, for the number of the request of 1 month right to use of 100MB network bandwidth resources, such as, have within first day 100 parts, there are 150 parts etc. within second day.

In actual applications, it is understood that there may be the request data of certain target resource is had no in the unit interval having, but When being tested with time series data, continuous data are generally required, so, this step can include：Obtaining history please Data are sought, and pending time series data is determined according to historical requests data；Process time sequence data is treated to be lacked The treatment of mistake value, determines time series data.Specifically, missing values treatment can include：Do not have when in certain unit interval During the request data of certain target resource, can by the value request in this unit interval and previous unit interval please Evaluation is consistent, or value request in latter unit interval is consistent, or the request in former and later two unit intervals The average value of value.

Step 12：According to time series data, time series trend information is determined, and it is true according to time series trend information The initial predicted value of the resource request that sets the goal.

Time series data is the basis as prediction, there is the information of tendency, can be true according to time series data Make the tendency information for time series.Specifically, time series data can be decomposed, determines that long-term trend are believed Breath, long-term trend can refer in longer time, a certain water persistently to be risen or fallen or rested on towards certain direction Variation tendency on flat, can be specifically by smooth value reflect from time series data origin-to-destination, with relative Consistent general trend.In actual applications, trend fractionation can be carried out by decompose function against time sequence data, Long-term trend information specifically can be obtained according to following formula：

F is odd number；

F is even number；

Wherein, t is the ordinal number of unit time, such as, the unit interval is day, in 1 year, first day t=1, second day t =2, etc..During X is historical requests data, actual request amount.L is length of time series, and f is period frequency.Such as, the time is worked as Sequence data when a length of 1 year when, l can take 365, if week for period frequency, f=7.

It is determined that during long-term trend information, it is possible to loss of learning occur, such as when f is 7, T₁、T₂、T₃、T₃₆₃、T₃₆₄ And T₃₆₅Equal no data, it is possible to by by T₁、T₂、T₃With T₄Be consistent, by T₃₆₃、T₃₆₄And T₃₆₅With T₃₆₂It is consistent Mode carries out completion.

After long-term trend information is determined, it is possible to using specific forecast model (or algorithm), determine target resource The initial predicted value of request.Such as, the initial predicted value of target resource request can by non-linear regression method, be determined.Also Moving average model can be integrated using autoregression, also known as arma modeling, determine the initial predicted value of target resource request.Specifically Ground, ARIMA models (p, d, q) can be expressed as follows：

Wherein, T_t' it is T_t, Δ^dT_t' it is T_t' sequence d jump sub-sequences, a_t, a_t-1... it is that the random of time series is disturbed Dynamic item, ψ₀、ψ₁……ψ_p, θ₁... ... θ_qEtc. being parameter to be estimated, generally parameter ψ is estimated with maximum-likelihood method₀、ψ₁……ψ_p, θ₁... ... θ_q.Disturbance refers to Δ^dT_t' subtract Δ^dT_t' predicted value value.

By T_tData input is to after forecast model, it is possible to obtain in following unit interval for resource request just Step predicted value.

In actual applications, only determine that although initial predicted value is feasible with long-term trend information, it is also possible to herein On the basis of further improve predicted value accuracy, in order to reach the more accurate purpose of predicted value, according to time series number According to, determining the long-term trend information of time series data, can also include according to time series data, determine time series data Long-term trend information, seasonal trend information and Stochastic Trends information.

Specifically, seasonal trend can refer to be showed regular due to being influenceed by certain fixed cycle sexual factor Cyclic fluctuation, such as, with the moon as fixed cycle, then the request amount in every month can be with regular fluctuation.In reality In the application of border, it is also possible to carry out trend fractionation by decompose function against time sequence data, decompose functions point Solution model has addition decomposition model (additive) and the decomposition model that is multiplied (multiplicative), specific as follows：

It is added decomposition model：

X_t=T_t+S_t+e_t

Multiplication decomposition model：

X_t=T_t×S_t×e_t

During X is historical requests data, actual request amount；T is long-term trend information.

As a example by being added decomposition model, seasonal trend information tool determines with according to following formula：

Wherein, n is rounded for l to f, and t%%f removes remainder to f for t, and when remainder is for 0, t%%f is f.

As can be seen that seasonal trend can determine that l is the time according to time series data and long-term trend information Sequence length, f is period frequency.

Stochastic Trends information can refer to uncertain factor, from being added decomposition model above：

e_t=X_t-T_t-S_t

After three kinds of tendency informations are determined, it is possible to be predicted, by the agency of, is carried out using ARIMA models above , in actual applications, with Stochastic Trends information can be added long-term trend information by prediction, in the lump as input data, utilize ARIMA models are predicted, and obtain the first predicted value, now T_t' can be just according to T_tWith e_tAddition is obtained.

The second predicted value can also be determined according to seasonal trend information, using third index flatness, specifically, Third index flatness, also known as Holt-Winters models, the seasonal trend information that will be determined by historical requests data As input information, using the model, the predicted value for resource request in following unit interval is determined.Specifically, may be used It is as follows with the model：

b_t=γ (S_t-S_t-1)+(1-γ)b_t-1

F_t+m=(S_t-b_tm)I_t-L+m

Wherein, S_tAs seasonal trend information, b_tIt is the Trend value in model, is intermediate parameters, I_tIt is for season is corrected Number, α, β, γ are parameter to be estimated, and are generally estimated using maximum-likelihood method.M is to being predicted week from a certain cycle in history The number of cycles of phase, F_t+mIt is the second predicted value determined.

After the first predicted value and the second predicted value is determined, it is possible to determine initial predicted value according to the two values.Tool Body ground, when Trend Decomposition is carried out to time series data, with addition decomposition model, it is determined that during initial predicted value, it is also possible to First predicted value is added with the second predicted value；When using multiplication decomposition model, it is determined that during initial predicted value, it is also possible to by first Predicted value is multiplied with the second predicted value.Step 13：Obtain the history exchange rate data of target resource and homegrown resource.

Because the exchange rate of target resource and homegrown resource can change, it is possible to go to obtain history exchange rate number According to history exchange rate data can also include some unit intervals, each unit interval correspondence one exchange rate, this step Unit interval can be consistent with the unit interval in historical requests data, such as, can be the unit time period with day.

Step 14：According to history exchange rate data, exchange rate tendency information is determined, and determine regulation system according to preset rules Number.

Can be predicted according to time series data in step 13, this step can also history exchange rate data to not The exchange rate come in unit interval is predicted.This step can include：According to history exchange rate data, using rolling average Line, determines the first exchange rate tendency information, and according to the first preset rules, determines the first adjustment factor；And/or handed over according to history Rate data are changed, using logarithm period power law model, the second exchange rate tendency information is determined, and according to the second preset rules, really Fixed second adjustment factor；According to the first adjustment factor and/or the second adjustment factor, adjustment factor is determined.

Specifically, Moving Average (Moving Average, MA) is generated by moving average, is by some lists Moving average in the time of position be connected as it is linear, so commonly referred to as Moving Average, referred to as equal line.If using day as Unit interval, then can have 5 average daily lines, 10 average daily lines etc..Several 5 daily means are exactly connected as line, are generated 5 Equal line.Specifically, moving average can be determined by following formula：

Wherein, MA_tIt is the moving average in t unit interval, n represents the cycle of being averaged, r_iIt is exchange rate.

Moving average can serve as the first law of communication tendency information, and according to the first preset rules, determine the first tune Section coefficient.First preset rules can be as follows：

When relatively short-term Moving Average is upward through longer-term Moving Average, can set predicted value adjustment factor is μ₁；

When the Moving Average of different cycles is upwardly extended, and equal line of short cycle is when upper, can set predicted value tune Section coefficient is μ₂；

When the Moving Average of different cycles is extended downwardly, and equal line of short cycle under when, can set predicted value tune Section coefficient is μ₃；

When relatively short-term Moving Average is passed down through longer-term Moving Average, can set predicted value adjustment factor is μ₄。

Wherein, μ₁And μ₂Could be arranged to more than 1, μ₃And μ₄Could be arranged to less than 1.

μ is the first adjustment factor.But the adjustment factor determined by Moving Average, is typically characterized in a short time Variation tendency.And logarithm period power law model can then characterize long-term variation tendency.

Logarithm period power law model (Log-Periodic Power Law models, LPPL models), can predict length The variation tendency of phase.Concrete model can have expressions below：

Wherein, r_tExchange rate during for unit time period t, A ＞ 0 are r_tIn the logarithm value of critical moment；B ＜ 0 are close for C R when 0_tThe logarithm value of increment in unit interval before moment t；C is the ratio of the fluctuation view picture of exponential increase The factor；t_c- t is the difference of unit time period t and critical value, and m is power exponent, and ω is foam vibration frequency；For phase is joined Number.Model parameter estimation aspect, first with least square method A, B, C t_c, m, ω andRepresent, then sought with genetic algorithm Look for t_c, m, ω andOptimal solution.

Finally, the second exchange rate tendency information B and t is determined_c.After obtaining the second exchange rate tendency information, it is possible to root According to the second preset rules, the second adjustment factor is determined.Second preset rules can be as follows：

As B ＜ 0, t_c＞ t or B ＞ 0, t_cDuring ＜ t, adjustment factor can be set for ρ₁；

As B ＜ 0, t_c＜ t or B ＞ 0, t_cDuring ＞ t, adjustment factor can be set for ρ₂；

ρ can be set₁＜ 1, ρ₂＞ 1, ρ are the second adjustment factor.

It is determined that during adjustment factor, can be determined according to the first adjustment factor and/or the second adjustment factor, above It has been noted that the first adjustment factor determined by Moving Average, typically characterizes variation tendency in a short time, and pass through The second regulation that logarithm period power law model is determined can then characterize long-term variation tendency.In order to reach preferably prediction Effect, makes adjustment factor more accurate, and the first adjustment factor and the second adjustment factor can be multiplied, and determines regulation system Number.

Step 15：According to initial predicted value and adjustment factor, the value request of target resource in prediction unit interval.

In step 13 and step 14, it has been determined that initial predicted value and adjustment factor, in this step, it is possible to Initial predicted value and adjustment factor are multiplied, the value request to target resource in unit interval is predicted, it is assumed that Initial predicted value is F₁, the value request of target resource is F in following unit interval₂, just there is following formula：

F₂=F₁×δ

Wherein, δ=μ, or δ=ρ, or δ=μ × ρ.

In actual applications, some special time periods, such as working time and time of having a rest are had, working day and section are false Day, these technical dates can also cause certain influence to the value request of target resource in unit interval, so, implement in one kind In mode, this step can include：According to initial predicted value, adjustment factor and special time period coefficient, the unit interval is predicted The value request of target resource in section.Specifically, can be determined by following formula：

F₂=F₁×δ×Q

Wherein, Q be special time period coefficient, such as, if with day be the unit time period, can be according to historical requests number According to it is determined that daily special time period coefficient.

In actual applications, there can be the value request of prediction in each unit interval, and during by each unit Between after section, can all have a value request for reality, so, in the case where data qualification is allowed, can according to multiple units when Between section interior prediction value request, and correspondence each unit interval in actual request value, by linear regression model (LRM), it is determined that Go out other regression coefficient, when being predicted to value request again, it is also possible to by regression coefficient and the unit interval for predicting The value request of target resource is multiplied in section, the value request of further Optimization Prediction.

Even if the accuracy in view of prediction is improved, but is eventually to predict and unreality, such as when with day as unit Between, in the value request of the target resource for predicting next day, and with homegrown resource exchange target resource after, arrived next day, Greatly it is possible that actual request value is with the value request predicted, there is any discrepancy.Based on the fact that, inventor is improve prediction While accuracy, it was also proposed that monitor in real time and the method for determining switch-activity.

In one embodiment, predicting in unit interval after the value request of the target resource, can also be such as There is following step shown in Fig. 3 and Fig. 4：

Step 16：According to the value request of target resource in the unit interval for predicting, target is exchanged by homegrown resource Resource.

In this step, exactly with the value request of the target resource in the unit interval that is predicted in step 15 be according to According to, with free resource go exchange target resource.

Step 17：When in unit interval, and reach the default period of the day from 11 p.m. to 1 a.m and carve, obtain by default period of the day from 11 p.m. to 1 a.m quarter target resource Real-time accumulated value request.

Here the unit interval refers to predict the corresponding unit interval in step 15, such as, be predicted within 15th in May, in advance The value request of target resource in May 16 is measured, and is exchanged with free resource, obtained the value request phase with prediction With target resource, then in unit interval can refer to just in May 16.

When default period of the day from 11 p.m. to 1 a.m quarter is reached, the real-time accumulated value request that target resource is carved by the default period of the day from 11 p.m. to 1 a.m is obtained.At one In unit interval, can there may be times when, such as be day when the unit interval, then in one day, have many period of the day from 11 p.m. to 1 a.m and carve, preset son Moment can be time point set in advance, can be random setting, or according to temporal regularity setting, such as, one day In, preset 4 sub- moment, 9:00、12:00、15:00, and 22:00；Can also be that each presets sub- time at intervals 1 hour, That can just have 11 sub- moment in one day.

The real-time accumulated value request for carving target resource by the default period of the day from 11 p.m. to 1 a.m refers to, since unit interval, to default son Moment, target resource adds up value request in real time.Such as, it is 9 to preset the period of the day from 11 p.m. to 1 a.m and carve:00, then on May 16, from 00:00 to 9: The real value request of 00 target resource in this 9 hours is exactly real-time accumulated value request.

Step 18：According to historical requests data, it is determined that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history.

In this step, the accumulative accounting of history that the period of the day from 11 p.m. to 1 a.m is carved is preset, refer in history unit interval, to be carved by the default period of the day from 11 p.m. to 1 a.m The ratio of the history of target resource accumulative value request value request in correspondence unit interval.Here historical requests data, can With identical with the time span of the historical requests data in step 11, it is also possible to different.

Specifically, can be determined by following formula：

Wherein, percent_itIt is the accumulative accounting of the history in t-th unit interval, being accumulated to the i moment；amt_itIt is t In individual unit interval, add up the accumulative value request of history at i moment；amt_tIt is the value request in t-th unit interval.Here The i moment, it is possible to be that the default period of the day from 11 p.m. to 1 a.m is carved.

Such as, on May 16, when arrival 9:When 00, it is possible in historical requests data, determine a certain in history In a few days, from 0:00 reaches 9:The accumulative value request of the history of 00 target resource, and in history this in a few days, amount to value request, By determining ratio, it is determined that being accumulated to 9:The accumulative accounting of 00 history.

Introduced in a step 11, missing values treatment was carried out to historical requests data, in step 17 and this step, also might be used So that this problem can be related to, when in historical requests data in the absence of the history that target resource is carved by the default period of the day from 11 p.m. to 1 a.m is accumulative please Evaluation, then be referred in previous unit interval, by the accumulative value request of history that the default period of the day from 11 p.m. to 1 a.m is carved, it is also possible to reference to same In one unit interval, by the accumulative value request of history that a default upper period of the day from 11 p.m. to 1 a.m carves target resource.Obtaining by the default period of the day from 11 p.m. to 1 a.m During the real-time accumulated value request carved, if there is non-existent situation, then can just be set to 0.

In actual applications, historical requests data are a lot, such as take the historical requests data of a year, just have 365 The individual default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, so, in one embodiment, in order to make full use of historical requests data, improve The accounting determined it is comprehensive, the default period of the day from 11 p.m. to 1 a.m in this step carves the accumulative accounting of history, can be multiple history unit interval In section, the accumulative value request of history for carving target resource by the default period of the day from 11 p.m. to 1 a.m in correspondence unit interval the ratio of value request it is flat Average.

Such as, historical requests data be 1 year, and with day be the unit time period, then just have 365 default period of the day from 11 p.m. to 1 a.m The accumulative accounting of history is carved, now, it is possible to average, an average default period of the day from 11 p.m. to 1 a.m is obtained and is carved the accumulative accounting of history.

In actual applications, certain unit interval has the unit interval with identical characteristics therewith, such as, with day It it is the unit time period, then if residing unit interval is Wednesday, then the Wednesday in historical requests data just has more The same sex, so, in one embodiment, the default period of the day from 11 p.m. to 1 a.m in this step carves the accumulative accounting of history, can also be with it is residing Unit interval has in the unit interval of identical characteristics, and the accumulative value request of history for carving target resource by the default period of the day from 11 p.m. to 1 a.m exists The ratio of value request in correspondence unit interval.

Such as, if residing unit interval is Wednesday, then the default period of the day from 11 p.m. to 1 a.m is carved the accumulative accounting of history and is just referred to The data of individual Wednesday.Or with reference to 4 nearest data of Wednesday, and be averaged, obtain the default period of the day from 11 p.m. to 1 a.m and carve the accumulative accounting of history.

Step 19：According to real-time progressive value request and the accumulative accounting of predetermined time, mesh in the residing unit interval of prediction Mark the value request of resource.

Real-time progressive value request is obtained in step 17, and defines the accumulative accounting of predetermined time in step 18, Just the value request in residing unit interval can be predicted according to the two values.

Specifically, can be by following formula：

Wherein, ftramt_tValue request in as residing unit interval.

Such as, when arrival 9:When 00, define and be accumulated to 9:The accumulative accounting of 00 history, and it is accumulative to have got the same day To 9:00 real-time progressive value request, it is possible to which the value request to the same day is predicted.

Step:110：According to the value request for predicting target resource in unit interval, during with the residing unit for predicting Between in section target resource value request, determine switch-activity.

By the agency of above, even if which kind of degree is the accuracy of prediction bring up to, is eventually prediction and unreality, and pole It is big that it is possible that actual request amount is with the request amount predicted, there is any discrepancy.So, this step just can be by according to real-time accumulated Value request in value request, and the residing unit interval that goes out of historical requests data prediction, and according only to historical requests data The value request in unit interval for predicting, is compared determination switch-activity, such as when difference is 5%, determine target The exchanged form of resource and homegrown resource.

Since can be according to real-time accumulated value request, and the request in the unit interval that goes out of historical requests data prediction Value, also just can predict the real-time accumulated request by any instant when in unit interval.So

Step 111：According to historical requests data, it is determined that default next period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history.

Similar with the accumulative accounting of default period of the day from 11 p.m. to 1 a.m quarter history, it can refers to just history to preset next period of the day from 11 p.m. to 1 a.m and carve the accumulative accounting of history In unit interval, the accumulative value request of history for carving target resource by default next period of the day from 11 p.m. to 1 a.m is asked in correspondence unit interval The ratio of value.

Such as, when arrival 9:When 00, it can be just 10 to preset next period of the day from 11 p.m. to 1 a.m and carve:00 (had every 1 hour one it is default when Carve), then can just be determined according to historical requests data, in history, by 10:The accumulative accounting of 00 history.

percent_jtCan think in t-th unit interval, be accumulated to the accumulative accounting of history at j moment, here the j moment It can be just next predetermined time at i moment.

Such as, there is a predetermined time every 1 hour, then i ∈ [1,11], j ∈ [i+1,11].

Step 112：According to the value request of target resource in the residing unit interval for predicting, and preset next period of the day from 11 p.m. to 1 a.m The accumulative accounting of history is carved, the value request of target resource is carved in prediction by default next period of the day from 11 p.m. to 1 a.m.

Specifically, can be by following formula predictions：

ftramt_jt=ftramt_t×percent_jt

Wherein, ftramt_jtThe value request carved by default next period of the day from 11 p.m. to 1 a.m in as residing unit interval.

Step 113：When default next period of the day from 11 p.m. to 1 a.m quarter is reached, obtain and carve tiring out in real time for target resource by default next period of the day from 11 p.m. to 1 a.m Meter value request.

Step 114：According to the value request that target resource is carved by default next period of the day from 11 p.m. to 1 a.m of prediction, with acquisition by default The real-time accumulated value request that next period of the day from 11 p.m. to 1 a.m is carved, determines switch-activity.

Such as, when reaching 9:When 00, step 23 has been predicted by 10:00 value request, then when reaching 10:00 When, it is possible to obtain by 10:00 real-time accumulated value request, and being compared according to prediction, determine switch-activity.

The method provided using embodiment 1, by the time series data determined according to historical requests data, it is determined that Time series trend information, and the initial predicted value of target resource request is determined, and by history exchange rate data, it is determined that Go out adjustment factor, further according to initial predicted value and adjustment factor, the value request to target resource is predicted, and solves existing skill The artificial subjective judgement of art causes the relatively low problem of accuracy, improves the accuracy of prediction target resource value request.In addition, passing through Real-time estimate simultaneously monitors residing discrepancy between the value request of time period and the value of prediction, determines switch-activity, and by reality When predict and monitor the value request and actual accumulative value request for being accumulated to predetermined time, determine switch-activity, can be right in time The homegrown resource and target resource for possessing are adjusted, and improve the utilization rate of resource.

Embodiment 2

Based on identical inventive concept, embodiment 2 provides a kind of prediction meanss of Internet resources value request, for carrying The accuracy of height prediction Internet resources value request.The prediction of the Internet resources value request that Fig. 5 is provided for the embodiment of the present application The structured flowchart of device, the device includes：First acquisition unit 21, the first determining unit 22, second acquisition unit 23, second are true Order unit 24 and predicting unit 25, wherein,

First acquisition unit 21, can be used for obtaining historical requests data, and determine the time according to historical requests data Sequence data, includes in time series data：The value request of target resource in some unit intervals；

First determining unit 22, can be used for according to time series data, determine time series trend information, and according to when Between Sequence Trend information determine target resource request initial predicted value；

Second acquisition unit 23, can be used for obtaining the history exchange rate data of target resource and homegrown resource；

Second determining unit 24, can be used for, according to history exchange rate data, determining exchange rate tendency information, and according to pre- If rule determines adjustment factor；

Predicting unit 25, can be used for according to initial predicted value and adjustment factor, target money in prediction unit interval The value request in source.

In one embodiment, the first determining unit 22, can be used for：

According to time series data, the long-term trend information of time series data is determined；

According to long-term trend information, using autoregressive moving-average model, the initial predicted value of target resource request is determined.

In one embodiment, the first determining unit 22, can be used for：

According to time series data, determine the long-term trend information of time series data, seasonal trend information and with Machine tendency information；

According to long-term trend information and Stochastic Trends information, using autoregressive moving-average model, the first prediction is determined Value；

According to seasonal trend information, using third index flatness, the second predicted value is determined；

According to the first predicted value and the second predicted value, the initial predicted value of target resource request is determined.

In one embodiment, the second determining unit 24, can be used for：

According to history exchange rate data, using moving average, the first exchange rate tendency information is determined, and it is pre- according to first If regular, the first adjustment factor is determined；And/or

According to history exchange rate data, using logarithm period power law model, the second exchange rate tendency information, and root are determined According to the second preset rules, the second adjustment factor is determined；

According to the first adjustment factor and/or the second adjustment factor, adjustment factor is determined.

In one embodiment, predicting unit 25, specifically for：

According to initial predicted value, adjustment factor and special time period coefficient, target resource in prediction unit interval Value request.

In one embodiment, first acquisition unit 21, can be used for：

Historical requests data are obtained, and pending time series data is determined according to historical requests data；

Treating process time sequence data carries out missing values treatment, determines time series data.

In one embodiment, the device also includes：Monitoring unit, can be used for：

According to the value request of target resource in the unit interval for predicting, target resource is exchanged by homegrown resource；

When in unit interval, and reach the default period of the day from 11 p.m. to 1 a.m and carve, obtain and carve the real-time of target resource by the default period of the day from 11 p.m. to 1 a.m Accumulative value request；

According to historical requests data, it is determined that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the accumulative accounting of history that the period of the day from 11 p.m. to 1 a.m is carved is preset, For in history unit interval, the accumulative value request of history for carving target resource by the default period of the day from 11 p.m. to 1 a.m please in correspondence unit interval The ratio of evaluation；

According to real-time progressive value request and the accumulative accounting of predetermined time, the value request in the residing unit interval of prediction；

According to mesh in the value request of target resource in the unit interval for predicting, with the residing unit interval for predicting The value request of resource is marked, switch-activity is determined.

In one embodiment, monitoring unit, can be used for：

According to historical requests data, it is determined that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved adds up Accounting is that in multiple history unit intervals, the accumulative value request of target resource history carved by the default period of the day from 11 p.m. to 1 a.m is single in correspondence The average value of the ratio of value request in the time period of position.

In one embodiment, monitoring unit, can be used for：

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved Accumulative accounting is have in the unit interval of identical characteristics with residing unit interval, and target money is carved by the default period of the day from 11 p.m. to 1 a.m The ratio of the history in source accumulative value request value request in correspondence unit interval.

In one embodiment, the monitoring unit, can be also used for：

According to historical requests data, it is determined that default next period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, what default next period of the day from 11 p.m. to 1 a.m was carved The accumulative accounting of history is in history unit interval, by the accumulative value request of history that default next period of the day from 11 p.m. to 1 a.m carves target resource The ratio of value request in correspondence unit interval；

Carve and go through according to the value request of target resource in the residing unit interval for predicting, and default next period of the day from 11 p.m. to 1 a.m History adds up accounting, and the value request of target resource is carved in prediction by default next period of the day from 11 p.m. to 1 a.m；

When default next period of the day from 11 p.m. to 1 a.m quarter is reached, obtain and carve tiring out in real time for target resource by default next period of the day from 11 p.m. to 1 a.m Meter value request；

According to the value request that target resource is carved by default next period of the day from 11 p.m. to 1 a.m of prediction, preset by described with obtaining Next period of the day from 11 p.m. to 1 a.m carves the real-time accumulated value request of target resource, determines switch-activity.

Embodiment 3

Based on identical invention thinking, as extension, a kind of Forecasting Methodology of internet business amount is present embodiments provided, Fund more and more circulates as a kind of internet information resource in internet, especially, with internet borderlessization, Many business of importation are poured in the country.This has occurred as soon as business of importation needs with the situation of the foreign exchange settlement.The exchange rate is not Disconnected change, for different revaluation and situation about devaluing, if it is possible to be predicted in advance, the requirement of anticipated redemption trading volume Currency, can not only effectively hide exchange rate risk, and can when fund be fully used.But prior art is people Work subjective judgement, such as, judge that RMB is converted dollar and to be devalued, and judges that tomorrow dollar transactions amount can be bigger than today, institute So that RMB is converted into dollar in advance, to tackle the transaction of tomorrow, but human subjective judges after all " eye is limited ", to lose Accuracy.So a kind of Forecasting Methodology of internet business amount is present embodiments provided, for improving prediction internet business amount Accuracy.The idiographic flow schematic diagram of the method is as shown in fig. 6, comprise the steps：

Step 31：Historical trading data is obtained, and time series data is determined according to historical trading data.

It is similar with step 11, historical trading data can be first obtained, a plurality of transaction record is included in historical trading data, often Can be comprising time, ID, trading value of target currency etc. in bar transaction record.So can be by all historical tradings Data are arranged sequentially in time, determine time series data, wherein target currency trading volume can be included.

In actual applications, it is also possible to the transaction data of target currency is had no in the unit interval that there are,

Included in time series data：Target currency trading volume in some unit intervals；So, this step can be wrapped Include：Historical trading data is obtained, and pending time series data is determined according to historical trading data；Treat process time sequence Column data carries out missing values treatment, determines time series data.

Step 32：According to time series data, time series trend information is determined, and it is true according to time series trend information The initial predicted value of the moneytary operations amount that sets the goal.

It is similar with step 12, it is also possible to according to time series data, long-term trend information is determined, recycle ARMA moulds Type, determines initial predicted value.

Or, according to time series data, determine long-term trend information, seasonal trend information, and Stochastic Trends Information, and according to arma modeling, and Holt-Winters models, is determined with predicted value and the second predicted value, then root Initial predicted value is determined according to the two predicted values.

Step 33：Obtain historical rate data.

The historical rate data of (such as 1 year) can be obtained in passing a period of time, such as, if own currency is the people Coin, target currency is dollar (or yen), it is necessary to obtain the data that dollar (or yen) in a year converts RMB exchange rate.

Step 34：According to historical rate data, exchange rate tendency information is determined, and determine adjustment factor according to preset rules.

After historical rate data are got, it is possible to similar with step 13, by moving average and/or logarithm Periodicity power law model, determines the first adjustment factor and/or the second adjustment factor.

When the first adjustment factor is determined by moving average, the first preset rules can be extended below (with the people As a example by coin is for own currency)：

When relatively short-term Moving Average is upward through longer-term Moving Average, it is believed that RMB devaluation possibility It is very big；

When the Moving Average of different cycles is upwardly extended, and equal line of short cycle is when upper, it is believed that RMB is demoted Value possibility is larger；

When the Moving Average of different cycles is extended downwardly, and equal line of short cycle under when, it is believed that RMB liter Value possibility is larger；

When relatively short-term Moving Average is passed down through longer-term Moving Average, it is believed that appreciation of the RMB possibility It is very big.

When the second adjustment factor is determined by property one number time power law model, the second preset rules can be extended below (still so that RMB is own currency as an example)：

As B ＜ 0, t_c＞ t or B ＞ 0, t_cDuring ＜ t, it is believed that RMB will be chronically at the trend of appreciation；

As B ＜ 0, t_c＜ t or B ＞ 0, t_cDuring ＞ t, it is believed that RMB will be chronically at the trend of devaluation.

Step 35：According to initial predicted value and adjustment factor, the trading volume of target currency in prediction unit interval.

Similar with step 15, this step can also include：It is according to initial predicted value, adjustment factor and special time period Number, the trading volume of target currency in prediction unit interval.

It is similar to Example 1, while forecasting accuracy is improve, it was also proposed that monitor in real time simultaneously determines exchange behavior Method, predicting in unit interval after the trading volume of target currency, can also be as shown in Figure 7 have following step：

Step 36：According to the trading volume of target currency in the unit interval for predicting, by own currency conversion target Currency.

Such as, predicting next day of trade will have 1,000,000 dollars of trading volume, it is possible to used corresponding people's currency exchange Go out this 1,000,000 dollars.

Step 37：When in unit interval, and reach the default period of the day from 11 p.m. to 1 a.m and carve, obtain by default period of the day from 11 p.m. to 1 a.m quarter target currency Real-time accumulated trading volume.

Step 38：According to historical trading data, it is determined that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history.

Step 39：According to real-time progressive trading volume and the accumulative accounting of predetermined time, mesh in the residing unit interval of prediction Mark the trading volume of currency.

Step 310：According to target currency trading volume in the unit interval for predicting, with the residing unit interval for predicting The transaction of target currency in section, it is determined that the behavior of exchange.

Step 311：According to historical trading data, it is determined that default next period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history.

Step 312：According to the trading volume of target currency in the residing unit interval for predicting, and preset next period of the day from 11 p.m. to 1 a.m The accumulative accounting of history is carved, the trading volume of target currency is carved in prediction by default next period of the day from 11 p.m. to 1 a.m.

Step 313：When default next period of the day from 11 p.m. to 1 a.m quarter is reached, obtain and carve tiring out in real time for target currency by default next period of the day from 11 p.m. to 1 a.m Meter trading volume.

Step 314：According to the trading volume that target currency is carved by default next period of the day from 11 p.m. to 1 a.m of prediction, with acquisition by default The real-time accumulated trading volume that next period of the day from 11 p.m. to 1 a.m is carved, it is determined that the behavior of exchange.

Because step 37 to step 314 is similar with step 17 step 114 in embodiment 1, repeat no more.

The method provided using embodiment 3, by the time series data determined according to historical trading data, it is determined that Time series trend information, and the initial predicted value of target currency trading volume is determined, and by historical rate data, it is determined that Go out adjustment factor, further according to initial predicted value and adjustment factor, the trading volume to target currency is predicted, and solves existing skill The artificial subjective judgement of art causes the relatively low problem of accuracy, improves the accuracy of prediction target currency trading volume.In addition, passing through Real-time estimate simultaneously monitors residing discrepancy between the trading volume of time period and the trading volume of prediction, it is determined that the behavior of exchange, Yi Jitong Cross real-time estimate and monitor the trading volume for being accumulated to predetermined time and actual accumulative trading volume, it is determined that the behavior of exchange, can be timely Ground is adjusted to the own currency and target currency that possess, improves the utilization rate of resource, also can to a certain extent hide remittance Rate risk.

It should be understood by those skilled in the art that, embodiments of the invention can be provided as method, system or computer program Product.Therefore, the present invention can be using the reality in terms of complete hardware embodiment, complete software embodiment or combination software and hardware Apply the form of example.And, the present invention can be used and wherein include the computer of computer usable program code at one or more The computer program implemented in usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) is produced The form of product.

The present invention is the flow with reference to method according to embodiments of the present invention, equipment (system) and computer program product Figure and/or block diagram are described.It should be understood that every first-class during flow chart and/or block diagram can be realized by computer program instructions The combination of flow and/or square frame in journey and/or square frame and flow chart and/or block diagram.These computer programs can be provided The processor of all-purpose computer, special-purpose computer, Embedded Processor or other programmable data processing devices is instructed to produce A raw machine so that produced for reality by the instruction of computer or the computing device of other programmable data processing devices The device of the function of being specified in present one flow of flow chart or multiple one square frame of flow and/or block diagram or multiple square frames.

These computer program instructions may be alternatively stored in can guide computer or other programmable data processing devices with spy In determining the computer-readable memory that mode works so that instruction of the storage in the computer-readable memory is produced and include finger Make the manufacture of device, the command device realize in one flow of flow chart or multiple one square frame of flow and/or block diagram or The function of being specified in multiple square frames.

These computer program instructions can be also loaded into computer or other programmable data processing devices so that in meter Series of operation steps is performed on calculation machine or other programmable devices to produce computer implemented treatment, so as in computer or The instruction performed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one The step of function of being specified in individual square frame or multiple square frames.

In a typical configuration, computing device includes one or more processors (CPU), input/output interface, net Network interface and internal memory.

Internal memory potentially includes the volatile memory in computer-readable medium, random access memory (RAM) and/or The forms such as Nonvolatile memory, such as read-only storage (ROM) or flash memory (flash RAM).Internal memory is computer-readable medium Example.

Computer-readable medium includes that permanent and non-permanent, removable and non-removable media can be by any method Or technology realizes information Store.Information can be computer-readable instruction, data structure, the module of program or other data. The example of the storage medium of computer includes, but are not limited to phase transition internal memory (PRAM), static RAM (SRAM), moves State random access memory (DRAM), other kinds of random access memory (RAM), read-only storage (ROM), electric erasable Programmable read only memory (EEPROM), fast flash memory bank or other memory techniques, read-only optical disc read-only storage (CD-ROM), Digital versatile disc (DVD) or other optical storages, magnetic cassette tape, the storage of tape magnetic rigid disk or other magnetic storage apparatus Or any other non-transmission medium, can be used to store the information that can be accessed by a computing device.Defined according to herein, calculated Machine computer-readable recording medium does not include temporary computer readable media (transitory media), such as data-signal and carrier wave of modulation.

Also, it should be noted that term " including ", "comprising" or its any other variant be intended to nonexcludability Comprising so that process, method, commodity or equipment including a series of key elements not only include those key elements, but also wrapping Include other key elements being not expressly set out, or also include for this process, method, commodity or equipment is intrinsic wants Element.In the absence of more restrictions, the key element limited by sentence "including a ...", it is not excluded that wanted including described Also there is other identical element in process, method, commodity or the equipment of element.

It will be understood by those skilled in the art that embodiments herein can be provided as method, system or computer program product. Therefore, the application can be using the embodiment in terms of complete hardware embodiment, complete software embodiment or combination software and hardware Form.And, the application can be used to be can use in one or more computers for wherein including computer usable program code and deposited The shape of the computer program product implemented on storage media (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) Formula.

Embodiments herein is the foregoing is only, the application is not limited to.For those skilled in the art For, the application can have various modifications and variations.It is all any modifications made within spirit herein and principle, equivalent Replace, improve etc., within the scope of should be included in claims hereof.

Claims

1. a kind of Forecasting Methodology of Internet resources value request, it is characterised in that including：

Historical requests data are obtained, and time series data, the time series number are determined according to the historical requests data Included in：The value request of target resource in some unit intervals；

According to the time series data, time series trend information is determined, and determine according to the time series trend information The initial predicted value of target resource request；

According to the initial predicted value and the adjustment factor, the value request of the target resource in prediction unit interval.

2. the method for claim 1, it is characterised in that according to the time series data, determine time series trend Information, and the initial predicted value that target resource is asked is determined according to the time series trend information, including：

According to the long-term trend information, using autoregressive moving-average model, the preliminary pre- of the target resource request is determined Measured value.

3. method as claimed in claim 2, it is characterised in that according to the time series data, determine the time series The long-term trend information of data, including：

According to the time series data, determine the long-term trend information of the time series data, seasonal trend information with And Stochastic Trends information；Then

According to the long-term trend information, using autoregressive moving-average model, the preliminary pre- of the target resource request is determined Measured value, including：

According to the long-term trend information and Stochastic Trends information, using autoregressive moving-average model, the first prediction is determined Value；

According to first predicted value and second predicted value, the initial predicted value of the target resource request is determined.

4. the method for claim 1, it is characterised in that according to the history exchange rate data, determine exchange rate trend Information, and determine adjustment factor according to preset rules, including：

According to the history exchange rate data, using moving average, the first exchange rate tendency information is determined, and it is pre- according to first If regular, the first adjustment factor is determined；And/or

According to the history exchange rate data, using logarithm period power law model, the second exchange rate tendency information, and root are determined According to the second preset rules, the second adjustment factor is determined；

5. the method for claim 1, it is characterised in that according to the initial predicted value and the adjustment factor, really The final predicted value of the fixed target resource request, including：

According to the initial predicted value, the adjustment factor and special time period coefficient, the mesh in prediction unit interval Mark the value request of resource.

6. the method for claim 1, it is characterised in that obtain historical data, and determined according to the historical data Time series data, including：

7. the method for claim 1, it is characterised in that methods described also includes：

According to the value request of the target resource in the unit interval for predicting, target resource is exchanged by homegrown resource；

When in the unit interval, and reach the default period of the day from 11 p.m. to 1 a.m and carve, obtain and carve the target money by the default period of the day from 11 p.m. to 1 a.m The real-time accumulated value request in source；

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved adds up Accounting is that in history unit interval, the accumulative value request of history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m is single in correspondence The ratio of value request in the time period of position；

According to the real-time progressive value request and the accumulative accounting of the predetermined time, the request in the residing unit interval of prediction Value；

According to the value request of the target resource in the unit interval for predicting, with the residing unit interval for predicting The value request of the interior target resource, determines switch-activity.

8. method as claimed in claim 7, it is characterised in that the accumulative accounting of history that the default period of the day from 11 p.m. to 1 a.m is carved is multiple history In unit interval, the history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m adds up value request in correspondence unit interval The average value of the ratio of value request.

9. method as claimed in claim 7, it is characterised in that the accumulative accounting of history that the default period of the day from 11 p.m. to 1 a.m is carved is and residing list The position time period has in the unit interval of identical characteristics, and the history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m adds up please The ratio of evaluation value request in correspondence unit interval.

10. method as claimed in claim 7, it is characterised in that methods described also includes：

According to historical requests data, it is determined that default next period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that default next period of the day from 11 p.m. to 1 a.m is carved Accumulative accounting is in history unit interval, by the accumulative value request of history that default next period of the day from 11 p.m. to 1 a.m carves the target resource The ratio of value request in correspondence unit interval；

Carve and go through according to the value request of the target resource in the residing unit interval for predicting, and default next period of the day from 11 p.m. to 1 a.m History adds up accounting, and the value request of the target resource is carved in prediction by default next period of the day from 11 p.m. to 1 a.m；

When default next period of the day from 11 p.m. to 1 a.m quarter is reached, obtain and carve tiring out in real time for the target resource by default next period of the day from 11 p.m. to 1 a.m Meter value request；

According to the value request that the target resource is carved by default next period of the day from 11 p.m. to 1 a.m of prediction, preset by described with obtaining Next period of the day from 11 p.m. to 1 a.m carves the real-time accumulated value request of the target resource, determines switch-activity.

A kind of 11. prediction meanss of Internet resources value request, it is characterised in that including：First acquisition unit, first determine list Unit, second acquisition unit, the second determining unit and predicting unit, wherein,

The first acquisition unit, for obtaining historical requests data, and determines time sequence according to the historical requests data Column data, includes in the time series data：The value request of target resource in some unit intervals；

First determining unit, for according to the time series data, determining time series trend information, and according to described Time series trend information determines the initial predicted value of target resource request；

Second determining unit, for according to the history exchange rate data, determining exchange rate tendency information, and according to default Rule determines adjustment factor；

The predicting unit, it is described in prediction unit interval for according to the initial predicted value and the adjustment factor The value request of target resource.

12. devices as claimed in claim 11, it is characterised in that first determining unit, specifically for：

13. devices as claimed in claim 12, it is characterised in that first determining unit, specifically for：

According to the time series data, determine the long-term trend information of the time series data, seasonal trend information with And Stochastic Trends information；

14. devices as claimed in claim 11, it is characterised in that second determining unit, specifically for：

15. devices as claimed in claim 11, it is characterised in that the predicting unit, specifically for：

16. devices as claimed in claim 11, it is characterised in that the first acquisition unit, specifically for：

17. devices as claimed in claim 11, it is characterised in that described device also includes：Monitoring unit, specifically for：

18. devices as claimed in claim 17, it is characterised in that the monitoring unit, specifically for：

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved adds up Accounting is that in multiple history unit intervals, the history for carving the target resource by the default period of the day from 11 p.m. to 1 a.m adds up value request right Answer the average value of the ratio of value request in unit interval.

19. devices as claimed in claim 17, it is characterised in that the monitoring unit, specifically for：

According to historical requests data, determine that the default period of the day from 11 p.m. to 1 a.m carves the accumulative accounting of history, the history that the default period of the day from 11 p.m. to 1 a.m is carved adds up Accounting is have in the unit interval of identical characteristics with residing unit interval, and the target money is carved by the default period of the day from 11 p.m. to 1 a.m The ratio of the history in source accumulative value request value request in correspondence unit interval.

20. devices as claimed in claim 17, it is characterised in that the monitoring unit, are additionally operable to：

A kind of 21. Forecasting Methodologies of internet business amount, it is characterised in that including：

Historical trading data is obtained, and time series data, the time series number are determined according to the historical trading data Included in：Target currency trading volume in some unit intervals；

According to the time series data, time series trend information is determined, and determine according to the time series trend information The initial predicted value of target currency trading volume；

Obtain historical rate data；

According to the initial predicted value and the adjustment factor, the trading volume of the target currency in prediction unit interval.