CN106875027B

CN106875027B - Resource request value prediction method and device, and transaction amount prediction method

Info

Publication number: CN106875027B
Application number: CN201610393294.8A
Authority: CN
Inventors: 喻继银; 潘晓峰
Original assignee: Advanced New Technologies Co 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: 2020-11-06
Anticipated expiration: 2036-06-06
Also published as: CN106875027A

Abstract

The application discloses a method and a device for predicting a requested value of an internet resource, which are used for improving the accuracy of predicting the requested value of the internet resource, and the method comprises the following steps: acquiring historical request data, and determining time sequence data according to the historical request data, wherein the time sequence data comprises: a request value of a target resource in a plurality of unit time periods; determining time series trend information according to the time series data, and determining a preliminary predicted value of the target resource request according to the time series trend information; acquiring historical exchange rate data of the target resource and the own resource; determining exchange rate trend information according to the historical exchange rate data, and determining an adjusting coefficient according to a preset rule; and predicting the request value of the target resource in unit time period according to the preliminary predicted value and the adjusting coefficient. The application also discloses a method for predicting the internet transaction amount.

Description

Resource request value prediction method and device, and transaction amount prediction method

Technical Field

The present disclosure relates to the field of internet information technologies, and in particular, to a method and an apparatus for predicting a requested value of an internet resource, and a method for predicting an internet transaction amount.

Background

With the development of internet information technology, various internet resources (such as network storage resources, network bandwidth resources, network computing resources, and the like, for short, resources) appear. Based on the "resource" property, it is usually necessary to avoid long time idle, and both the own party and the other party should use the resource as much as possible, so as to maximize the efficiency of the resource. In the internet, multiple parties can own different own resources, but the own resources are not necessarily owned but are needed, so that the needs can be acquired respectively by means of resource exchange. For example, a certain amount of network storage resources is used to replace a certain amount of network bandwidth resources. Resource management systems for managing resources have thus emerged. The system can manage various resources and exchange target resources by self resources, after receiving the resource request of the user, if the system self owns the resources corresponding to the request, the system can directly provide the self resources to the user, and if the system self does not own, the system can exchange the target resources corresponding to the request through the self resources and provide the target resources to the user.

Because the resources are various in types and the supply and demand relationship of different types of resources in the internet is also changed, a certain exchange rate occurs according to the market factors such as the supply and demand relationship during resource exchange, for example, 1 month usage right of 1TB network storage resources can be exchanged with 1 month usage right of 100MB network bandwidth resources, but if idle network storage resources in the internet are increased, the network storage resources are over-supplied at this time, and it is possible that 1 month usage right of 1.5TB network storage resources is needed if 1 month usage right of 100MB network bandwidth resources is exchanged again. In the "internet resource market", the exchange rate of various resources may change at any time. Based on the unstable characteristic of the resource exchange rate, the resource management system may exchange in advance according to the requirement of the user for the resource, that is, exchange the own resource as the target resource in advance, and after receiving the resource request of the user, the resource management system may directly provide the target resource corresponding to the request to the user.

In the prior art, when resources are exchanged in advance, the exchanged resource values are only obtained by manually judging the change condition of the exchange rate and the change condition of a user request value, and each target resource request value and the exchange rate trend of the next unit time period or next unit time periods are subjectively determined, so that the resources are exchanged in advance. However, it is obvious that the manual prediction of the resource exchange rate and the target resource request value is inaccurate.

Disclosure of Invention

The embodiment of the application provides a method for predicting a requested value of an internet resource, which is used for improving the accuracy of predicting the requested value of the internet resource.

The embodiment of the application provides a device for predicting a requested value of an internet resource, which is used for improving the accuracy of predicting the requested value of the internet resource.

The embodiment of the application provides a method for predicting internet transaction amount, which is used for improving the accuracy of predicting the internet transaction amount.

The embodiment of the application adopts the following technical scheme:

a method for predicting a requested value of an Internet resource comprises the following steps:

acquiring historical request data, and determining time sequence data according to the historical request data, wherein the time sequence data comprises: a request value of a target resource in a plurality of unit time periods;

determining time series trend information according to the time series data, and determining a preliminary predicted value of the target resource request according to the time series trend information;

acquiring historical exchange rate data of the target resource and the own resource;

determining exchange rate trend information according to the historical exchange rate data, and determining an adjusting coefficient according to a preset rule;

and predicting the request value of the target resource in unit time period according to the preliminary predicted value and the adjusting coefficient.

Preferably, determining time series trend information according to the time series data, and determining a preliminary predicted value of the target resource request according to the time series trend information comprises:

determining long-term trend information of the time sequence data according to the time sequence data;

and determining a preliminary predicted value of the target resource request by utilizing an autoregressive moving average model according to the long-term trend information.

Preferably, determining long-term trend information of the time series data according to the time series data comprises:

determining long-term trend information, seasonal trend information and random trend information of the time sequence data according to the time sequence data; then

Determining a preliminary predicted value of the target resource request by using an autoregressive moving average model according to the long-term trend information, wherein the preliminary predicted value comprises the following steps:

determining a first predicted value by utilizing an autoregressive moving average model according to the long-term trend information and the random trend information;

determining a second predicted value by utilizing a cubic exponential smoothing method according to the seasonal trend information;

and determining a preliminary predicted value of the target resource request according to the first predicted value and the second predicted value.

Preferably, determining exchange rate trend information according to the historical exchange rate data, and determining an adjustment coefficient according to a preset rule, includes:

determining first exchange rate trend information by utilizing a moving average value according to the historical exchange rate data, and determining a first adjusting coefficient according to a first preset rule; and/or

Determining second exchange rate trend information by utilizing a log periodic power law model according to the historical exchange rate data, and determining a second adjusting coefficient according to a second preset rule;

and determining an adjusting coefficient according to the first adjusting coefficient and/or the second adjusting coefficient.

Preferably, determining a final predicted value of the target resource request according to the preliminary predicted value and the adjustment coefficient includes:

and predicting the request value of the target resource in unit time period according to the preliminary predicted value, the adjusting coefficient and the special time period coefficient.

Preferably, acquiring historical data and determining time series data according to the historical data includes:

acquiring historical request data, and determining time sequence data to be processed according to the historical request data;

and performing missing value processing on the time sequence data to be processed to determine the time sequence data.

Preferably, the method further comprises:

exchanging the target resource through the owned resource according to the predicted request value of the target resource in the unit time period;

when the time is in the unit time period and reaches a preset sub-moment, acquiring a real-time accumulated request value of the target resource up to the preset sub-moment;

determining the historical accumulated proportion of the preset sub-moment according to historical request data, wherein the historical accumulated proportion of the preset sub-moment is the ratio of the historical accumulated request value of the target resource in the corresponding unit time period when the preset sub-moment is reached;

predicting the request value in the unit time period according to the real-time accumulated request value and the accumulated ratio at the preset moment;

and determining the switching behavior according to the predicted request value of the target resource in the unit time period and the predicted request value of the target resource in the unit time period.

Preferably, the historical accumulation ratio of the preset sub-time is an average value of ratios of the historical accumulation request values of the target resource in the corresponding unit time periods within a plurality of historical unit time periods up to the preset sub-time.

Preferably, the historical accumulation ratio of the preset sub-time is a ratio of the historical accumulation request value of the target resource to the request value of the corresponding unit time period within the unit time period having the same characteristic as the unit time period.

Preferably, the method further comprises:

determining a preset next sub-moment historical accumulated proportion according to historical request data, wherein the preset next sub-moment historical accumulated proportion is the ratio of the historical accumulated request value of the target resource in the corresponding unit time period within the historical unit time period until the preset next sub-moment;

predicting the request value of the target resource up to the preset next sub-moment according to the predicted request value of the target resource in the unit time period and the historical accumulated ratio of the preset next sub-moment;

when the preset next sub-moment is reached, acquiring a real-time accumulated request value of the target resource until the preset next sub-moment;

and determining the switching behavior according to the predicted request value of the target resource up to the preset next sub-moment and the acquired real-time accumulated request value of the target resource up to the preset next sub-moment.

An apparatus for predicting a requested value of an internet resource, the apparatus comprising: a first obtaining unit, a first determining unit, a second obtaining unit, a second determining unit, and a predicting unit, wherein,

the first obtaining unit is configured to obtain history request data, and determine time series data according to the history request data, where the time series data includes: a request value of a target resource in a plurality of unit time periods;

the first determining unit is used for determining time series trend information according to the time series data and determining a preliminary predicted value of the target resource request according to the time series trend information;

the second obtaining unit is used for obtaining historical exchange rate data of the target resource and the own resource;

the second determining unit is used for determining exchange rate trend information according to the historical exchange rate data and determining an adjusting coefficient according to a preset rule;

and the prediction unit is used for predicting the request value of the target resource in a unit time period according to the preliminary prediction value and the regulation coefficient.

Preferably, the first determining unit is specifically configured to:

determining long-term trend information, seasonal trend information and random trend information of the time sequence data according to the time sequence data;

Preferably, the second determining unit is specifically configured to:

Preferably, the prediction unit is specifically configured to:

Preferably, the first obtaining unit is specifically configured to:

Preferably, the apparatus further comprises: the monitoring unit is specifically used for:

Preferably, the monitoring unit is specifically configured to:

and determining the historical accumulated proportion of the preset sub-moment according to historical request data, wherein the historical accumulated proportion of the preset sub-moment is the average value of the ratio of the historical accumulated request value of the target resource in the corresponding unit time period within a plurality of historical unit time periods up to the preset sub-moment.

Preferably, the monitoring unit is specifically configured to:

and determining the historical accumulated proportion of the preset sub-moment according to the historical request data, wherein the historical accumulated proportion of the preset sub-moment is the ratio of the historical accumulated request value of the target resource in the corresponding unit time period within the unit time period with the same characteristic as the unit time period.

Preferably, the monitoring unit is further configured to:

A method for predicting an internet transaction amount, comprising:

acquiring historical transaction data, and determining time sequence data according to the historical transaction data, wherein the time sequence data comprises: a target currency transaction amount over a number of unit time periods;

determining time series trend information according to the time series data, and determining a preliminary predicted value of the target currency transaction amount according to the time series trend information;

obtaining historical exchange rate data;

and predicting the transaction amount of the target currency in a unit time period according to the preliminary predicted value and the adjusting coefficient.

The embodiment of the application adopts at least one technical scheme which can achieve the following beneficial effects: the time sequence trend information is determined according to the time sequence data determined by the historical request data, the preliminary predicted value of the target resource request is determined, the adjusting coefficient is determined according to the historical exchange rate data, and the requested value of the target resource is predicted according to the preliminary predicted value and the adjusting coefficient, so that the problem of low accuracy caused by artificial subjective judgment in the prior art is solved, and the accuracy of predicting the requested value of the target resource is improved. In addition, the exchange behavior is determined by predicting and monitoring the request value and the predicted value in the time period in real time, and the exchange behavior is determined by predicting and monitoring the request value and the actual accumulated request value accumulated to the preset moment in real time, so that the owned self-resources and the owned target resources can be adjusted in time, and the utilization rate of the resources is improved.

Drawings

The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:

fig. 1 is a schematic flowchart of a method for predicting a requested value of an internet resource according to embodiment 1 of the present application;

fig. 2 is a schematic diagram illustrating a method for predicting a requested value of an internet resource according to embodiment 1 of the present application;

fig. 3 is a schematic flowchart of a method for monitoring and determining exchange behavior in real time according to embodiment 1 of the present application;

FIG. 4 is a schematic diagram of real-time prediction of a target resource request value provided in embodiment 1 of the present application;

fig. 5 is a block diagram illustrating a structure of an apparatus for predicting a requested value of an internet resource according to embodiment 2 of the present application;

fig. 6 is a flowchart illustrating a method for predicting an internet transaction amount according to embodiment 3 of the present application;

fig. 7 is a flowchart illustrating a method for monitoring and determining redemption behavior in real time according to embodiment 3 of the present application.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be described in detail and completely with reference to the following specific embodiments of the present application and the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The technical solutions provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings.

Example 1

As described above, the prior art predicts the resource exchange rate and the target resource request value based on artificial subjective judgment, for example, if the user wants to exchange the 1-month usage right of 100MB network bandwidth resources, the user needs to exchange the network storage resources with larger and larger capacity of the 1-month usage right, and the request value of the network bandwidth resources is more and more, so the network bandwidth resources can be exchanged in advance, and the requirement of the user on the network bandwidth resources is met, but the accuracy is obviously lower in an artificial manner. Therefore, based on the defect, the inventor provides a method for predicting a requested value of an internet resource, which is used for improving the accuracy of predicting the requested value of the internet resource, and the specific flow diagram of the method is shown in fig. 1 and fig. 2, and the method comprises the following steps:

step 11: and acquiring historical request data, and determining time series data according to the historical request data.

When internet resource request values are predicted, historical data are needed to be used as a basis, so the step can be used for obtaining the historical request data, for example, all the data requesting the internet resources in the past year can have records containing time, user identification and request values of target resources at every moment. Therefore, in this step, different target resources can be divided into different time-series data. For example, 1 month usage right for 100MB network bandwidth resource, there may be a time series data, which may include: if the requested value for the target resource is set to a number of times per unit time, for example, the time unit is set to day, then the time series data may be the number of requested copies of the 1-month usage right for 100MB of the network bandwidth resource, for example, 100 copies on the first day, 150 copies on the second day, etc.

In practical applications, there may be some request data that does not have a certain target resource in a unit time period, but when testing with time series data, continuous data is often needed, so this step may include: acquiring historical request data, and determining time sequence data to be processed according to the historical request data; and carrying out missing value processing on the time sequence data to be processed to determine the time sequence data. Specifically, the missing value processing may include: when there is no request data of a certain target resource in a certain unit time period, the request value in the unit time period may be kept consistent with the request value in the previous unit time period, or the request value in the next unit time period may be kept consistent, or the average value of the request values in the previous unit time period and the next unit time period may be kept.

Step 12: and determining time series trend information according to the time series data, and determining a preliminary predicted value of the target resource request according to the time series trend information.

The time-series data is information having a tendency as a basis of prediction, and the tendency information for the time-series can be determined from the time-series data. Specifically, the time series data may be decomposed to determine long-term trend information, where the long-term trend may be a trend that continuously rises or falls in a certain direction or stays at a certain level for a long time, and may be a relatively consistent overall trend from a start point to an end point of the time series data reflected by a smooth value. In practical application, the trend of the time series data can be split through a decompose function, and the long-term trend information can be obtained according to the following formula:

f is an odd number;

f is an even number;

where t is the ordinal number of the unit time, for example, the unit time is day, and in one year, t =1 on the first day, t =2 on the second day, etc. X is the actual request amount in the historical request data. l is the time series length and f is the cycle frequency. For example, when the time-series data has a duration of 1 year, l may be 365, and if the cycle frequency is week, f = 7.

When determining long-term trend information, information loss may occur, for example, when f is 7, T₁、T₂、T₃、T₃₆₃、T₃₆₄And T₃₆₅All have no data, so can pass T₁、T₂、T₃And T₄Keep consistent, will T₃₆₃、T₃₆₄And T₃₆₅And T₃₆₂In a consistent mannerAnd (5) completing.

After the long-term trend information is determined, a specific prediction model (or algorithm) can be used to determine a preliminary prediction value of the target resource request. For example, the preliminary prediction value of the target resource request may be determined by a non-linear regression method. An autoregressive integral moving average model, also known as an ARMA model, may also be utilized to determine a preliminary prediction value for a target resource request. Specifically, the ARIMA model (p, d, q) can be expressed as follows:

wherein, T_tIs' T_t，

Is T_t' d-order differential sequences of sequences, a_t，a_t-1… … is a time series of random perturbation terms, ψ₀、ψ₁……ψ_p，θ₁，……θ_qWaiting for the parameter to be estimated, the parameter psi is usually estimated using maximum likelihood₀、ψ₁……ψ_p，θ₁，……θ_q. The random disturbance term is Δ^dT_t' minus delta^dT_t' value of the predicted value.

At the time of T_tAfter the data is input into the prediction model, a preliminary prediction value for the resource request in a unit time period in the future can be obtained.

In practical application, although it is feasible to determine the preliminary predicted value by using only the long-term trend information, the accuracy of the predicted value can be further improved on the basis of the preliminary predicted value, and in order to achieve the purpose of more accurate predicted value, the long-term trend information of the time series data is determined according to the time series data, and the long-term trend information, the seasonal trend information and the random trend information of the time series data can be determined according to the time series data.

Specifically, a seasonal trend may refer to a periodic fluctuation that appears regularly due to some fixed periodic factor, for example, a fixed period of months, so that the request amount per month may fluctuate regularly. In practical application, the time series data may also be trend-split by a decomplex function, and the decomposition models of the decomplex function include an additive decomposition model (additive) and a multiplicative decomposition model (multiplicative), which are specifically as follows:

additive decomposition model:

multiplication decomposition model:

x is the actual request quantity in the historical request data; and T is long-term trend information.

Taking the additive decomposition model as an example, the seasonal trend information is determined according to the following formula:

where n is l rounded to f, t%% f is t to f remainder, and when the remainder is 0, t%% f is f.

It can be seen that the seasonal trend may be determined from the time series data and the long term trend information, i being the length of the time series and f being the cycle frequency.

The random trend information may refer to uncertain factors, as known from the additive decomposition model above:

after the three trend information are determined, prediction can be performed, as described above, prediction is performed by using the ARIMA model, in practical application, long-term trend information and random trend information can be added together to be used as input data, prediction is performed by using the ARIMA model to obtain a first predicted value, and at this moment, T is_t' may be according to T_tAnd e_tAnd adding the two to obtain the final product.

The second predicted value can be determined by utilizing a cubic exponential smoothing method according to the seasonal trend information, specifically, the cubic exponential smoothing method, also called Holt-Winters model, takes the seasonal trend information determined by the historical request data as input information, and determines the predicted value for the resource request in the future unit time period by utilizing the model. Specifically, the model may be as follows:

wherein S is_tI.e. seasonal trend information, b_tIs the trend value in the model, is the intermediate parameter, I_tFor the seasonal correction coefficient, α, β, γ are parameters to be estimated, and are usually estimated by a maximum likelihood method. m is the number of cycles from a certain cycle in the history to the cycle to be predicted, F_t+mThe second predicted value is determined.

When the first predicted value and the second predicted value are determined, the preliminary predicted value can be determined according to the first predicted value and the second predicted value. Specifically, when performing trend decomposition on the time series data, the first predicted value and the second predicted value may also be added when determining the preliminary predicted value by using the additive decomposition model; when a multiplicative decomposition model is used, the first predicted value and the second predicted value may also be multiplied when determining the preliminary predicted value.

Step 13: and acquiring historical exchange rate data of the target resource and the own resource.

Since the exchange rate between the target resource and the own resource is variable, the historical exchange rate data may be obtained, the historical exchange rate data may also include a plurality of unit time periods, each unit time period corresponds to one exchange rate, and the unit time periods in this step may be consistent with the unit time periods in the historical request data, for example, all may take days as the unit time periods.

Step 14: and determining exchange rate trend information according to the historical exchange rate data, and determining an adjusting coefficient according to a preset rule.

In step 13, prediction can be performed based on the time-series data, and in this step, the exchange rate per unit time period in the future can be predicted based on the historical exchange rate data. This step may include: determining first exchange rate trend information by utilizing a moving average line according to historical exchange rate data, and determining a first adjusting coefficient according to a first preset rule; and/or determining second exchange rate trend information by using a log periodic power law model according to historical exchange rate data, and determining a second adjusting coefficient according to a second preset rule; and determining an adjusting coefficient according to the first adjusting coefficient and/or the second adjusting coefficient.

Specifically, the Moving Average (MA) is generated by Moving averages, and the Moving averages in a unit time are connected to form a linear shape, and therefore, the Moving Average is generally called a Moving Average, which is simply called an Average. If the unit time period is day, there may be a 5-day average line, a 10-day average line, etc. Several 5-day averages are connected as a line to generate a 5-day average. Specifically, the moving average may be determined by the following formula:

wherein, MA_tIs a moving average value in t unit time periods, n represents an averaged period, r_iIs the exchange rate.

The moving average value can be used as first commutative law trend information, and a first adjusting coefficient is determined according to a first preset rule. The first preset rule may be as follows:

when the shorter term moving average line passes upward through the longer term moving average line, the predictor adjustment coefficient may be set to μ₁；

When the moving average lines of different periods all extend upwards and the short period average line is up, the predicted value adjusting coefficient can be set to be mu₂；

When the moving average lines of different periods all extend downwards and the short period average line is at the bottom, the predicted value adjusting coefficient can be set to be mu₃；

When the shorter term moving average line passes down through the longer term moving average line, the predictor adjustment coefficient may be set to μ₄。

Wherein, mu₁And mu₂Can be set to be larger than 1, mu₃And mu₄May be set to less than 1.

Mu is the first regulating coefficient. But the adjustment coefficients, determined by moving the mean line, generally characterize the trend of change over the short term. And the log periodic power law model can represent the long-term variation trend.

The long-term change trend can be predicted by a Log-Periodic Power Law model (LPPL model). The specific model may have the following expression:

wherein r is_tIs the exchange rate per unit time period t, A>0 is r_tA logarithmic value at a critical time; b is<R when 0 is C close to 0_tA logarithmic value of the increase in unit time before the time t; c is a scale factor of the whole fluctuation of exponential growth; t is t_c-t is monoThe difference between the bit time period t and a critical value, m is a power exponent, and omega is a foam fluctuation frequency; phi is more than or equal to 0 and less than or equal to 2 pi is a phase parameter. In the aspect of model parameter estimation, A, B, C is firstly expressed as t by using a least square method_cM, omega and phi, and then using a genetic algorithm to find t_cM, ω and φ.

Finally, second exchange rate trend information B and t are determined_c. After the second exchange rate trend information is obtained, a second adjustment coefficient can be determined according to a second preset rule. The second preset rule may be as follows:

when B is present<0，t_c>t or B>0，t_c<At t, the adjustment coefficient may be set to ρ₁；

When B is present<0，t_c<t or B>0，t_c>At t, the adjustment coefficient may be set to ρ₂;

Can set ρ₁<1，ρ₂>1, rho is the second regulating coefficient.

In determining the adjustment factor, the determination may be performed based on the first adjustment factor and/or the second adjustment factor, and it has been mentioned above that the first adjustment factor determined by moving the mean line generally represents the short-term variation trend, and the second adjustment determined by the log-periodic power law model may represent the long-term variation trend. In order to achieve a better prediction effect and make the adjustment coefficient more accurate, the first adjustment coefficient and the second adjustment coefficient may be multiplied to determine the adjustment coefficient.

Step 15: and predicting the request value of the target resource in the unit time period according to the preliminary predicted value and the adjusting coefficient.

Having determined the preliminary prediction value and the adjustment factor in

steps

12 and 14, the preliminary prediction value and the adjustment factor may be multiplied in this step to predict the requested value of the target resource per unit time period, assuming that the preliminary prediction value is F₁The requested value of the target resource in the unit time period in the future is F₂There is the following equation:

wherein the content of the first and second substances,

or is or

Or is or

。

In practical applications, there may be some special time periods, such as working time and rest time, working day and holiday, and these special dates will also have certain influence on the requested value of the target resource in a unit time period, so in an embodiment, this step may include: and predicting the request value of the target resource in the unit time period according to the preliminary predicted value, the adjusting coefficient and the special time period coefficient. Specifically, it can be determined by the following formula:

wherein Q is a special time period coefficient, for example, if the time period is a unit of day, the special time period coefficient of each day can be determined according to the historical request data.

In practical application, each unit time period can have a predicted request value, and after each unit time period, an actual request value exists, so that under the condition that data conditions allow, another regression coefficient can be determined through a linear regression model according to the predicted request values in a plurality of unit time periods and the actual request value corresponding to each unit time period, and when the request value is predicted again, the regression coefficient can be multiplied by the predicted request value of the target resource in the unit time period, so that the predicted request value is further optimized.

Considering that the prediction is not realistic even if the accuracy of the prediction is improved, for example, a requested value of a target resource on the next day is predicted in a unit time of day, and the target resource is exchanged with an own resource, and then, the actual requested value and the predicted requested value are likely to be in or out. Based on this fact, the inventor also provides a method for monitoring and determining the exchange behavior in real time while improving the prediction accuracy.

In one embodiment, after predicting the requested value of the target resource in the unit time period, the following steps may be further performed as shown in fig. 3 and 4:

step 16: and exchanging the target resource through the owned resource according to the predicted request value of the target resource in the unit time period.

In this step, the own resource is used to exchange the target resource based on the requested value of the target resource in the unit time period predicted in step 15.

And step 17: and when the current time reaches the preset sub-moment within the unit time period, acquiring the real-time accumulated request value of the target resource up to the preset sub-moment.

The unit time refers to the unit time predicted in step 15, for example, if the unit time is predicted in 5 months and 15 days, the requested value of the target resource within 5 months and 16 days is predicted, and the target resource identical to the predicted requested value is obtained by exchanging the own resource, then the unit time within the unit time period may be 5 months and 16 days.

And when the preset sub-moment is reached, acquiring a real-time accumulated request value of the target resource up to the preset sub-moment. In a unit time, there are many moments, for example, when the unit time is a day, there are many sub-moments in the day, and the preset sub-moment may be a preset time point, may be set randomly, or may be set according to a time rule, for example, 4 sub-moments, 9:00, 12:00, 15:00, and 22:00, are preset in the day; or, each preset sub-moment may be spaced by 1 hour, and there may be 11 sub-moments in a day.

The real-time accumulated request value of the target resource up to the preset sub-moment means the real-time accumulated request value of the target resource from the unit time period to the preset sub-moment. For example, if the preset sub-time is 9:00, the real requested value of the target resource is the real-time accumulated requested value within 9 hours from 00:00 to 9:00 on day 5, month 16.

Step 18: and determining the historical accumulated ratio of the preset sub-moment according to the historical request data.

In this step, the historical accumulated percentage of the preset sub-time refers to a ratio of the historical accumulated requested value of the target resource to the requested value of the corresponding unit time period within the historical unit time period. The history request data may be the same as or different from the time span of the history request data in step 11.

Specifically, it can be determined by the following formula:

wherein, percent_itThe historical accumulated proportion accumulated to the moment i in the t unit time period; amt_itAccumulating the historical accumulated request value at the moment i in the t unit time period; amt_tIs the requested value in the t unit time period. Here, the time i may be a preset sub-time.

For example, on day 16/5, when 9:00 is reached, the history request data may specify the history accumulated request value for the 9:00 target resource from 0:00 on a certain day in the history, and the total request value on the certain day in the history, and specify the history accumulated occupancy ratio accumulated to 9:00 by specifying the ratio.

As described in step 11, the missing value processing is performed on the history request data, and this problem may be also involved in step 17 and this step, and when there is no history accumulated request value of the target resource up to the preset sub-time in the history request data, the history accumulated request value up to the preset sub-time in the previous unit time period may be referred to, or the history accumulated request value of the target resource up to the previous sub-time in the same unit time period may be referred to. When the real-time accumulation request value up to the preset sub-time is acquired, if the real-time accumulation request value does not exist, the real-time accumulation request value can be set to 0.

In practical applications, the historical request data is numerous, for example, when the historical request data of one year is taken, 365 preset sub-time historical accumulation ratios exist, so in one embodiment, in order to fully utilize the historical request data and improve the comprehensiveness of the determined ratio, the preset sub-time historical accumulation ratio in this step may be an average value of ratios of the historical accumulation request values of the target resource to the preset sub-time request values in corresponding unit time periods in a plurality of historical unit time periods.

For example, if the historical request data is a year and a day is a unit time period, 365 preset sub-time historical cumulative percentages exist, and at this time, an average value can be obtained to obtain an average preset sub-time historical cumulative percentage.

In practical applications, a certain unit time period has a unit time period with the same characteristics as the certain unit time period, for example, if the unit time period is a wednesday, then the wednesday in the history request data is more similar, so in one embodiment, the preset sub-time history accumulation ratio in this step may also be a ratio of the request value in the corresponding unit time period to the history accumulation request value of the target resource at the preset sub-time within the unit time period with the same characteristics as the unit time period.

For example, if the unit time period is wednesday, the preset sub-time historical cumulative percentage may refer to the last data of wednesday. Or referring to the latest 4 Wednesday data, and averaging to obtain the historical cumulative percentage of the preset sub-time.

Step 19: and predicting the request value of the target resource in the unit time period according to the real-time accumulated request value and the preset sub-moment historical accumulated ratio.

The real-time cumulative request value is obtained in step 17, and the historical cumulative percentage at the preset sub-time is determined in step 18, so that the request value in the unit time period can be predicted according to the two values.

Specifically, the following formula can be used:

wherein ftramt_tI.e. the requested value in the unit time period.

For example, when 9:00 is reached, the historical cumulative percentage accumulated to 9:00 is determined, and the real-time cumulative request value accumulated to 9:00 on the current day is acquired, so that the request value on the current day can be predicted.

Step 110: and determining the switching behavior according to the predicted request value of the target resource in the unit time period and the predicted request value of the target resource in the unit time period.

It has been described above that even to what extent the accuracy of the prediction is improved, it is predicted rather than realistic and it is highly likely that the actual request amount will deviate from the predicted request amount. Therefore, this step can determine the switching behavior by comparing the real-time accumulated request value and the predicted request value in the unit time period with the predicted request value in the unit time period based on the historical request data, for example, when the difference is 5%, determine the switching manner between the target resource and the owned resource.

Since the request value for the real-time accumulation can be predicted from the request value for the unit time period predicted from the request value for the real-time accumulation and the request value for the unit time period predicted from the request data for the history, the request value for the real-time accumulation up to any time can be predicted when the request value for the unit time period is within the unit time period. Therefore, it is not only easy to use

Step 111: and determining the preset historical accumulated occupation ratio of the next sub-moment according to the historical request data.

Similar to the preset sub-time historical cumulative percentage, the preset next sub-time historical cumulative percentage may be a ratio of the historical cumulative request value of the target resource to the corresponding unit time period within the historical unit time period.

For example, when 9:00 is reached, the preset next sub-time may be 10:00 (one preset time every 1 hour), and then it may be determined from the history request data that the cumulative percentage of the history is 10: 00.

percent_jtThe historical accumulated proportion of the accumulated time to the time j in the t unit time period can be obtained, and the time j can be the next preset time of the time i.

For example, every 1 hour there is a predetermined time, then i e [1,11], j e [ i +1,11 ].

Step 112: and predicting the target resource request value up to the preset next sub-moment according to the predicted target resource request value in the unit time period and the preset next sub-moment historical accumulated ratio.

Specifically, it can be predicted by the following formula:

wherein ftramt_jtNamely the requested value up to the preset next sub-moment in the unit time period.

Step 113: and when the preset next sub-moment is reached, acquiring the real-time accumulated request value of the target resource up to the preset next sub-moment.

Step 114: and determining the switching behavior according to the predicted request value of the target resource up to the preset next sub-moment and the acquired real-time accumulated request value up to the preset next sub-moment.

For example, when 9:00 is reached, step 23 has predicted the requested value up to 10:00, then when 10:00 is reached, the real-time accumulated requested value up to 10:00 may be obtained and compared against the predictions to determine the swap behavior.

By adopting the method provided by the embodiment 1, the time series trend information is determined according to the time series data determined by the historical request data, the preliminary predicted value of the target resource request is determined, the adjusting coefficient is determined according to the historical exchange rate data, and the requested value of the target resource is predicted according to the preliminary predicted value and the adjusting coefficient, so that the problem of low accuracy caused by artificial subjective judgment in the prior art is solved, and the accuracy of predicting the requested value of the target resource is improved. In addition, the exchange behavior is determined by predicting and monitoring the request value and the predicted value in the time period in real time, and the exchange behavior is determined by predicting and monitoring the request value and the actual accumulated request value accumulated to the preset moment in real time, so that the owned self-resources and the owned target resources can be adjusted in time, and the utilization rate of the resources is improved.

Example 2

Based on the same inventive concept, embodiment 2 provides a device for predicting a requested value of an internet resource, which is used for improving the accuracy of predicting the requested value of the internet resource. Fig. 5 is a block diagram illustrating a structure of an apparatus for predicting a requested value of an internet resource according to an embodiment of the present application, where the apparatus includes: a first acquisition unit 21, a first determination unit 22, a second acquisition unit 23, a second determination unit 24, and a prediction unit 25, wherein,

the first obtaining unit 21 may be configured to obtain history request data, and determine time series data according to the history request data, where the time series data includes: a request value of a target resource in a plurality of unit time periods;

a first determining unit 22, configured to determine time-series trend information according to the time-series data, and determine a preliminary predicted value of the target resource request according to the time-series trend information;

a second obtaining unit 23, configured to obtain historical exchange rate data between the target resource and the own resource;

the second determining unit 24 may be configured to determine exchange rate trend information according to the historical exchange rate data, and determine an adjustment coefficient according to a preset rule;

the prediction unit 25 may be configured to predict a requested value of the target resource in the unit time period based on the preliminary predicted value and the adjustment coefficient.

In an embodiment, the first determining unit 22 may be configured to:

determining long-term trend information of the time series data according to the time series data;

In an embodiment, the first determining unit 22 may be configured to:

In an embodiment, the second determining unit 24 may be configured to:

determining first exchange rate trend information by utilizing the moving average value according to historical exchange rate data, and determining a first adjusting coefficient according to a first preset rule; and/or

Determining second exchange rate trend information by using a log periodic power law model according to historical exchange rate data, and determining a second adjusting coefficient according to a second preset rule;

In one embodiment, the prediction unit 25 is specifically configured to:

and predicting the request value of the target resource in the unit time period according to the preliminary predicted value, the adjusting coefficient and the special time period coefficient.

In an embodiment, the first obtaining unit 21 may be configured to:

and carrying out missing value processing on the time sequence data to be processed to determine the time sequence data.

In one embodiment, the apparatus further comprises: a monitoring unit operable to:

when the time is in a unit time period and reaches a preset sub-moment, acquiring a real-time accumulated request value of a target resource up to the preset sub-moment;

determining a historical accumulated proportion of a preset sub-moment according to historical request data, wherein the historical accumulated proportion of the preset sub-moment is a ratio of a historical accumulated request value of a target resource intercepted to the preset sub-moment in a corresponding unit time period within a historical unit time period;

predicting the request value in the unit time period according to the real-time accumulated request value and the preset sub-moment historical accumulated ratio;

In one embodiment, the monitoring unit may be configured to:

and determining a historical accumulation ratio of a preset sub-moment according to historical request data, wherein the historical accumulation ratio of the preset sub-moment is an average value of ratios of the target resource historical accumulation request values up to the preset sub-moment in corresponding unit time periods in a plurality of historical unit time periods.

In one embodiment, the monitoring unit may be configured to:

and determining the historical accumulated proportion of the preset sub-moment according to the historical request data, wherein the historical accumulated proportion of the preset sub-moment is the ratio of the historical accumulated request value of the target resource to the request value of the preset sub-moment in the corresponding unit time period within the unit time period with the same characteristic as the unit time period.

In an embodiment, the monitoring unit may be further configured to:

determining a preset next sub-moment historical accumulated proportion according to historical request data, wherein the preset next sub-moment historical accumulated proportion is the ratio of the historical accumulated request value of the target resource at the preset next sub-moment to the request value in the corresponding unit time period within the historical unit time period;

when the preset next sub-moment is reached, acquiring a real-time accumulated request value of the target resource up to the preset next sub-moment;

and determining the switching behavior according to the predicted request value of the target resource at the next preset sub-moment and the acquired real-time accumulated request value of the target resource at the next preset sub-moment.

Example 3

Based on the same invention thought, as an extension, the embodiment provides a method for predicting the internet transaction amount, funds are used as an internet information resource and are increasingly circulated in the internet, and particularly, with the internet being unbounded, a lot of imported services are greatly brought into the country. This is the case when imported services need to be settled with foreign exchange. The exchange rate is constantly changing, and due to the conditions of different currencies of value increment and value decrement, if the currency required by the transaction amount can be converted in advance by predicting in advance, the risk of the exchange rate can be effectively avoided, and the fund can be fully utilized. However, the prior art is artificially subjective judgment, for example, the RMB is judged to be devalued and the daily dollar transaction amount is judged to be larger than that of the current day, so the RMB is converted into the dollar in advance to deal with the daily transaction, but the artificially subjective judgment is 'eye-limited' and has inaccuracy. Therefore, the embodiment provides a method for predicting the internet transaction amount, which is used for improving the accuracy of predicting the internet transaction amount. The specific flow diagram of the method is shown in fig. 6, and the method comprises the following steps:

step 31: and acquiring historical transaction data, and determining time sequence data according to the historical transaction data.

Similar to step 11, historical transaction data may be obtained, where the historical transaction data includes a plurality of transaction records, and each transaction record may include time, user identification, transaction value of the target currency, and the like. All historical transaction data may be arranged in chronological order to determine time series data, which may include the target currency transaction amount.

In practical applications, there may also be a unit of time during which there is no transaction data for the target currency,

the time-series data includes: a target currency transaction amount over a number of unit time periods; therefore, the step may include: acquiring historical transaction data, and determining time sequence data to be processed according to the historical transaction data; and carrying out missing value processing on the time sequence data to be processed to determine the time sequence data.

Step 32: time series trend information is determined from the time series data, and a preliminary predictive value of the target currency transaction amount is determined from the time series trend information.

Similar to step 12, long-term trend information can be determined according to the time series data, and then an ARMA model is utilized to determine a preliminary prediction value.

Or determining long-term trend information, seasonal trend information and random trend information according to the time sequence data, determining a first predicted value and a second predicted value according to an ARMA model and a Holt-Winters model, and determining a primary predicted value according to the two predicted values.

Step 33: historical exchange rate data is obtained.

Historical exchange rate data may be obtained over a period of time (e.g., a year), such as if the own currency is renminbi and the target currency is U.S. dollars (or yen), which may require a conversion of the renminbi to U.S. dollars (or yen) over the year.

Step 34: and determining exchange rate trend information according to the historical exchange rate data, and determining an adjusting coefficient according to a preset rule.

After obtaining the historical exchange rate data, the first adjustment coefficient and/or the second adjustment coefficient may be determined by a moving average and/or a log periodic power law model, similar to that in step 13.

In determining the first adjustment factor by means of the moving average, the first predetermined rule may be extended as follows (for example, renminbi is the self-owned currency):

when the shorter-term moving average line passes upward through the longer-term moving average line, it is considered that the renminbi depreciation possibility is high;

when the moving average lines of different periods all extend upwards and the short period average line is up, the possibility of the dereferencing of the RMB is considered to be high;

when the moving average lines of different periods all extend downwards and the short period average line is below, the possibility of renminbi increasing is considered to be high;

when the shorter term moving average is passed down the longer term moving average, it is considered that the probability of renminbi rising is high.

When the second adjustment coefficient is determined by the power-law model, the second predetermined rule may be extended as follows (the renminbi is still taken as the self-owned currency as an example):

when B is present<0，t_c>t or B>0，t_c<At t, the RMB is considered to be in the trend of increasing value for a long time;

when B is present<0，t_c<t or B>0，t_c>At t, it can be considered that the renminbi will be in a depreciation trend for a long time.

Step 35: and predicting the transaction amount of the target currency in the unit time period according to the preliminary predicted value and the adjusting coefficient.

Similar to step 15, this step may also include: and predicting the transaction amount of the target currency in the unit time period according to the preliminary predicted value, the adjusting coefficient and the special time period coefficient.

Similar to embodiment 1, while improving the prediction accuracy, a method for monitoring and determining the exchange behavior in real time is also proposed, and after predicting the transaction amount of the target currency in a unit time period, the following steps may be performed as shown in fig. 7:

step 36: and exchanging the target currency by the self-owned currency according to the predicted transaction amount of the target currency in the unit time period.

For example, if a transaction amount of 100 ten thousand dollars is predicted to be available on the next transaction day, the corresponding RMB can be exchanged for the 100 ten thousand dollars.

Step 37: and when the current time is within the unit time period and reaches the preset sub-moment, acquiring the real-time accumulated transaction amount of the target currency up to the preset sub-moment.

Step 38: and determining the historical accumulated ratio of the preset sub-moment according to the historical transaction data.

Step 39: and predicting the transaction amount of the target currency in the unit time period according to the real-time accumulated transaction amount and the preset sub-moment historical accumulated ratio.

Step 310: and determining the conversion behavior according to the predicted target currency transaction amount in the unit time period and the predicted target currency transaction in the unit time period.

Step 311: and determining the preset historical accumulated ratio of the next sub-moment according to the historical transaction data.

Step 312: and predicting the transaction amount of the target currency up to the preset next sub-moment according to the predicted transaction amount of the target currency in the unit time period and the preset next sub-moment historical accumulation ratio.

Step 313: and when the preset next sub-moment is reached, acquiring the real-time accumulated transaction amount of the target currency up to the preset next sub-moment.

Step 314: and determining the conversion behavior according to the predicted transaction amount of the target currency up to the preset next sub-moment and the acquired real-time accumulated transaction amount up to the preset next sub-moment.

Since steps 37 to 314 are similar to step 17 and step 114 in embodiment 1, they are not described again.

By adopting the method provided by the embodiment 3, the time series trend information is determined through the time series data determined according to the historical transaction data, the preliminary predicted value of the target currency transaction amount is determined, the adjusting coefficient is determined through the historical exchange rate data, and then the transaction amount of the target currency is predicted according to the preliminary predicted value and the adjusting coefficient, so that the problem of low accuracy caused by artificial subjective judgment in the prior art is solved, and the accuracy of predicting the transaction amount of the target currency is improved. In addition, the exchange behavior is determined by predicting and monitoring the transaction amount in the time period and the predicted transaction amount in real time, and the exchange behavior is determined by predicting and monitoring the transaction amount accumulated to the preset time and the actual accumulated transaction amount in real time, so that the owned currency and the target currency can be adjusted in time, the utilization rate of resources is improved, and the risk of exchange rate can be avoided to a certain extent.

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include forms of volatile memory in a computer readable medium, Random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for predicting a requested value of an Internet resource, comprising:

predicting the request value of the target resource in unit time period according to the preliminary predicted value and the adjusting coefficient;

wherein the target resource comprises a target currency, the owned resource comprises an owned currency, and the exchange rate comprises an exchange rate.

2. The method of claim 1, wherein determining time series trend information from the time series data and determining a preliminary predicted value of a target resource request from the time series trend information comprises:

3. The method of claim 2, wherein determining long-term trend information for the time series data from the time series data comprises:

4. The method of claim 1, wherein determining exchange rate trend information based on the historical exchange rate data and determining an adjustment factor based on a predetermined rule comprises:

5. The method of claim 1, wherein determining a final predicted value of the target resource request based on the preliminary predicted value and the adjustment factor comprises:

6. The method of claim 1, wherein obtaining historical data and determining time series data from the historical data comprises:

7. The method of claim 1, wherein the method further comprises:

predicting the requested value in the unit time period according to the real-time accumulated requested value and the historical accumulated ratio of the preset sub-moment;

8. The method as claimed in claim 7, wherein the historical accumulation ratio at the preset sub-time is an average value of ratios of the historical accumulation request values of the target resource to the corresponding unit time periods within a plurality of historical unit time periods.

9. The method as claimed in claim 7, wherein the historical accumulation ratio of the preset sub-time is a ratio of the request value of the historical accumulation request value of the target resource in the corresponding unit time period up to the preset sub-time within the unit time period having the same characteristic as the unit time period.

10. The method of claim 7, wherein the method further comprises:

11. An apparatus for predicting a requested value of an internet resource, comprising: a first obtaining unit, a first determining unit, a second obtaining unit, a second determining unit, and a predicting unit, wherein,

the prediction unit is used for predicting the request value of the target resource in a unit time period according to the preliminary prediction value and the regulation coefficient;

12. The apparatus of claim 11, wherein the first determining unit is specifically configured to:

13. The apparatus of claim 12, wherein the first determining unit is specifically configured to:

14. The apparatus of claim 11, wherein the second determining unit is specifically configured to:

15. The apparatus as claimed in claim 11, wherein said prediction unit is specifically configured to:

16. The apparatus of claim 11, wherein the first obtaining unit is specifically configured to:

17. The apparatus of claim 11, wherein the apparatus further comprises: the monitoring unit is specifically used for:

18. The apparatus according to claim 17, wherein the monitoring unit is specifically configured to:

19. The apparatus according to claim 17, wherein the monitoring unit is specifically configured to:

20. The apparatus of claim 17, wherein the monitoring unit is further configured to: