CN107729708A - A kind of traffic policy recommends method and device - Google Patents
A kind of traffic policy recommends method and device Download PDFInfo
- Publication number
- CN107729708A CN107729708A CN201610654128.9A CN201610654128A CN107729708A CN 107729708 A CN107729708 A CN 107729708A CN 201610654128 A CN201610654128 A CN 201610654128A CN 107729708 A CN107729708 A CN 107729708A
- Authority
- CN
- China
- Prior art keywords
- user
- flow
- month
- app
- kinds
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
- G16Z99/00—Subject matter not provided for in other main groups of this subclass
Abstract
The invention discloses a kind of traffic policy to recommend method, including：The service condition parameter that all kinds of APP are applied according to user, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；According to the basic use information parameter of user, second affecting parameters of the structure basic use information of user to customer flow demand；According to first affecting parameters, the second affecting parameters, the object function for predicting customer flow demand is established；Customer flow demand is predicted according to the object function, recommends traffic policy to user.Meanwhile the invention also discloses a kind of traffic policy recommendation apparatus.
Description
Technical field
The present invention relates to flow recommended technology, more particularly to a kind of traffic policy to recommend method and device.
Background technology
Flow operation, is the Main Management direction of operator's business in the mobile Internet epoch, and mobile service wound
The important means that new and storage client possesses.With the popularization of mobile Internet and the rapid development of big data, the application of flow
Scope has expanded to " shopping class ", " GT grand touring ", " financing class ", " social class " etc. now from initial " QQ ", " browser " etc.
Numerous mobile terminal APP applicating categories.The abundant and development of mobile terminal APP applications, improve the traffic demand of terminal user.
However, the various and complicated all kinds of APP of classification are using the influence to customer flow demand, recommend to operator to user rational
Flow package proposes challenge.
Existing flow package recommendation method, operator are generally used with the caller voice call duration and flow of user
Situation as point of penetration, returned using logistic, association analysis, decision tree scheduling algorithm, analyze the actual use of user itself
Matching between flow and existing flow package, by quantify the actual use flow of user itself and existing flow package it
Between " distance ", so as to recommend relatively reasonable flow package to user.However, two can so be present：One be by
In the changeful of internet and the wind vane of big data, and mobile terminal such as cell phone application application develops rapidly, traditional mould
Type can not accurately excavate the traffic demand of user from APP applications, so as to realize the lifting of flow package fit；Separately
One be due to conventional model be based on user's essential characteristic structure static models, it is impossible to implementation model dynamic tracking and
The adaptive tuning of model.
The content of the invention
In view of this, the embodiment of the present invention it is expected that providing a kind of traffic policy recommends method and device, can accurately know
The traffic demand of user, so as to recommend rational traffic policy to user.
To reach above-mentioned purpose, the technical proposal of the invention is realized in this way：
The embodiments of the invention provide a kind of traffic policy to recommend method, and methods described includes：
The service condition parameter that all kinds of APP are applied according to user, the service condition pair that structure user applies to all kinds of APP
First affecting parameters of customer flow demand；
According to the basic use information parameter of user, the structure basic use information of user influences on the second of customer flow demand
Parameter；
According to first affecting parameters, the second affecting parameters, the object function for predicting customer flow demand is established；
Customer flow demand is predicted according to the object function, recommends traffic policy to user.
In such scheme, methods described also includes：
According to the historical forecast effect of the object function, the object function is optimized.
In such scheme, the service condition parameter flows including monthly access times, monthly use duration, monthly use
Between amount, of that month and last month between access times amplitude of variation, this month and last month using duration amplitude of variation, it is of that month with last month it
Between use changes in flow rate amplitude；
The capital consumption information parameter comprises at least following a kind of：It is monthly to be made using telephone expenses, the of that month duration of call, this month
With each contact person in the quantity of the contact person in flow, of that month contact contacts list, contacts list when monthly average uses
The service condition parameter that each contact person applies to all kinds of APP in flow, contacts list.
In such scheme, the service condition parameter applied according to user to all kinds of APP, structure user is to all kinds of APP
The service condition of application to the first affecting parameters of customer flow demand, including：
The service condition parameter applied according to user to all kinds of APP, obtain the preference index that user applies to all kinds of APP；
The preference index includes preference weight amplitude of variation between preference weight, this month and last month；
Stream in the service condition parameter that the preference index applied according to user to all kinds of APP, user apply to all kinds of APP
Measure service condition parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；Institute
State flow service condition parameter and use changes in flow rate amplitude using between flow, this month and last month including monthly.
It is described to recommend traffic policy to user in such scheme, including：
If the traffic policy that the customer flow of the prediction is more than the of that month actual use flow of user and prepares recommendation
Expense is less than the expense for exceeding flow, then the traffic policy is recommended into the user；It is described to refer to institute beyond flow
State the difference of the customer flow and the actual use flow of prediction.
The embodiment of the present invention additionally provides a kind of traffic policy recommendation apparatus, and described device includes：Parameter structure module, mesh
Scalar functions establish module, recommending module；Wherein,
The parameter builds module, and for the service condition parameter applied according to user to all kinds of APP, structure user is to each
First affecting parameters of the service condition of class APP applications to customer flow demand；According to the basic use information parameter of user, structure
Second affecting parameters of the basic use information of user to customer flow demand；
The object function establishes module, for according to first affecting parameters, the second affecting parameters, establishing prediction and using
The object function of family traffic demand；
The recommending module, for predicting user's request flow according to the object function, recommend traffic policy to user.
In such scheme, described device also includes：Optimization module, for being imitated according to the historical forecast of the object function
Fruit, the object function is optimized.
In such scheme, the service condition parameter flows including monthly access times, monthly use duration, monthly use
Between amount, of that month and last month between access times amplitude of variation, this month and last month using duration amplitude of variation, it is of that month with last month it
Between use changes in flow rate amplitude；
The capital consumption information parameter comprises at least following a kind of：It is monthly to be made using telephone expenses, the of that month duration of call, this month
With each contact person in the quantity of the contact person in flow, of that month contact contacts list, contacts list when monthly average uses
The service condition parameter that each contact person applies to all kinds of APP in flow, contacts list.
In such scheme, the parameter builds module, is specifically used for：
The service condition parameter applied according to user to all kinds of APP, obtain the preference index that user applies to all kinds of APP；
The preference index includes preference weight amplitude of variation between preference weight, this month and last month；
Stream in the service condition parameter that the preference index applied according to user to all kinds of APP, user apply to all kinds of APP
Measure service condition parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；Institute
State flow service condition parameter and use changes in flow rate amplitude using between flow, this month and last month including monthly.
In such scheme, the recommending module, it is specifically used for：
If the traffic policy that the customer flow of the prediction is more than the of that month actual use flow of user and prepares recommendation
Expense is less than the expense for exceeding flow, then the traffic policy is recommended into the user；It is described to refer to institute beyond flow
State the difference of the customer flow and the actual use flow of prediction.
Traffic policy provided in an embodiment of the present invention recommends method and device, the use applied according to user to all kinds of APP
Situation parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；According to user
Basic use information parameter, second affecting parameters of the structure basic use information of user to customer flow demand；According to described
One affecting parameters, the second affecting parameters, establish the object function of prediction customer flow demand；Predicted and used according to the object function
Family traffic demand, recommend traffic policy to user；It can be seen that the embodiment of the present invention is made by analyzing user to all kinds of APP applications
With situation and the basic use information of user, the object function for predicting customer flow demand is established, so as to accurately know use
The traffic demand at family, recommend more reasonable, accurate traffic policy to user.
In addition, historical forecast effect of the embodiment of the present invention according to the object function, is carried out certainly to the object function
Adapt to adjustment so that the object function preferably can recommend rational traffic policy, lifting user's actual use to user
Fit between flow and the traffic policy for preparing recommendation.
Brief description of the drawings
Fig. 1 is the implementation process schematic diagram that traffic policy of the embodiment of the present invention recommends method；
Fig. 2 is the composition structural representation of traffic policy recommendation apparatus of the embodiment of the present invention.
Embodiment
A kind of traffic policy provided in an embodiment of the present invention recommends method, as shown in figure 1, methods described includes：
Step 101：The service condition parameter applied according to user to all kinds of APP, structure user make to all kinds of APP applications
The first affecting parameters with situation to customer flow demand；
Specifically, the service condition parameter first applied according to user to all kinds of APP, obtain what user applied to all kinds of APP
Preference index, the preference index include preference weight amplitude of variation between preference weight, this month and last month；Further according to user couple
Flow service condition parameter in the service condition parameter that the preference index of all kinds of APP applications, user apply to all kinds of APP, structure
Build first affecting parameters of the user to the service condition that all kinds of APP are applied to customer flow demand, the flow service condition ginseng
Number includes monthly use and uses changes in flow rate amplitude between flow, this month and last month.
Here, the user includes to the service condition parameter that all kinds of APP are applied：When monthly access times, monthly use
Long, monthly use changes width between access times amplitude of variation, this month and last month between flow, this month and last month using duration
Changes in flow rate amplitude is used between degree, of that month and last month；
Wherein, the user refers to user within continuous N number of moon to such to every class APP monthly access times applied
The average access times of APP applications, the monthly use duration that the user applies to every class APP refer to user in continuous N number of moon
Interior to use duration to the average of such APP applications, the monthly use flow that the user applies to every class APP refers to user even
Continue in N number of moon and flow, N >=2 are used to the average of such APP applications；
The APP applications can be divided into 30 classifications, including social activity, Online Video, short-sighted frequency, game, music, reading, hand
Machine browser, Domestic News, ecommerce, fresh electric business, bank, tourism trip, financing, map, the trip of city bus, love and marriage
Friend-making, image beautification, cloud disk, mobile phone safe, cuisines, physical culture, automobile, household house ornamentation, house property, medical treatment ＆ health, body beautification body-building, female
Property, mother and baby, education, life weather.
Here, in being applied according to user to above-mentioned 30 class APP when the monthly access times of every class APP applications, monthly use
Long, monthly use changes width between access times amplitude of variation, this month and last month between flow, this month and last month using duration
Using 6 indexs such as changes in flow rate amplitudes between degree, of that month and last month, calculate what user applied to every class APP based on Information Entropy
Preference weight amplitude of variation, is described in detail below between preference weight, this month and last month：
Remember pi,1The monthly access times applied for user i to such APP, pi,2Such APP is applied for user i monthly
Use duration, pi,3Monthly flow, p are used for what user i applied to such APPi,4This month applied for user i to such APP
The access times amplitude of variation between last month, pi,5Become between last month for user i this month applied to such APP and using duration
Change amplitude, pi,6For user i this month applied to such APP and use changes in flow rate amplitude between last month.
1) according to formulaCalculate entropy H of the user to every class APP each indexs appliedk, whereinM represents number of users；
Here, to make the data p of different indexsi,kHomogeneity, can be to the data p of all different indexsi,kAdvanced rower is accurate
Change is handled, so as to facilitate subsequent treatment.
2) according to formulaCalculate entropy weight w of the user to every class APP each indexs appliedk；
3) according to formulaCalculate the preference ω ' that user applies to every class APPj, j=1,2 ..., 30；
4) according to formulaThe preference ω ' applied to the user to every class APPjIt is normalized, obtains
Take the preference weight ω that family is applied to every class APPj；
5) according to formula △ ωj=ωj(T)-ωj(T-1)Calculate preference between this month and last month that user applies to every class APP
Weight amplitude of variation △ ωj, wherein, T represents of that month, and T-1 represents the of that month last month.
Here, the preference weight amplitude of variation △ ωjUser can be embodied between of that month and last month to such APP
Using the variation tendency of preference, △ ωjIt is bigger, then it represents that user is bigger using the change of preference to such APP, right
The influence of customer flow demand is also more notable；△ωjIt is smaller, then it represents that user gets over to such APP using the change of preference
Small, the influence to customer flow demand is also got over not notable.
Here, it is represented by according to above-mentioned condition, first affecting parameters to customer flow demand
Wherein, f (ωj,△ωj) represent the preference weight ω that user applied in this month to jth class APPjAnd it is corresponding of that month with last month it
Between preference weight amplitude of variation △ ωjAdjustment weight parameter under collective effect；βjWhat expression was applied to jth class APP, which estimates, is
Number, can be configured according to user to every class APP actual use situations applied；φ(xh) represent that user applies to jth class APP
Flow service condition parameter xhInfluence subparameter to customer flow demand, the flow service condition parameter xhIncluding monthly
Using using changes in flow rate amplitude between flow, this month and last month；H represents the ginseng included in the flow service condition parameter
Several numbers.
Step 102：According to the basic use information parameter of user, the structure basic use information of user is to customer flow demand
Second affecting parameters；
Here, the basic use information parameter of the user comprises at least following a kind of：Sex, age, occupation, monthly use
Telephone expenses, the of that month duration of call, the quantity of the of that month contact person using in flow, of that month contact contacts list, contacts list
In each contact person each contact person in monthly average is using flow, contacts list service condition ginseng that all kinds of APP applys
Number.
Here, second affecting parameters to customer flow demand are represented byWherein, βnExpression pair
Parameter x included in the basic use information parameter of usernCorresponding estimates coefficient, can use letter substantially according to user
The actual conditions of breath are configured；Represent the parameter x included in the basic use information parameter of the usernUser is flowed
The influence subparameter of amount demand；R represents the number of parameters included in the basic use information parameter of user.
Here, each contact person is every in monthly average is using flow, contacts list in the contacts list of the user
Individual contact person may influence the flow use to the user itself on the service condition parameter that all kinds of APP are applied, and therefore, need
Build second affecting parameters of the basic use information of user to customer flow demand.
Step 103：According to first affecting parameters, the second affecting parameters, the target for predicting customer flow demand is established
Function；
Here, according to the user obtained in step 101 to the service condition that all kinds of APP are applied to customer flow need
The basic use information of the user obtained in the first affecting parameters and step 102 asked is to the of customer flow demand
Two affecting parameters, using adaptive optimal Support vector regression (Self-adaptive Optimization Support
Vector Regression, SAO-SVR) algorithm establish prediction customer flow demand object function f (x), it is as follows：
Wherein, b is constant term.
Step 104：Customer flow demand is predicted according to the object function, recommends traffic policy to user.
Specifically, customer flow demand is predicted according to the object function f (x) obtained in step 103, if the prediction
Customer flow is more than the of that month actual use flow of user and the expense for the traffic policy for preparing to recommend exceeds flow less than described
Expense, then the traffic policy is recommended into the user；It is described to refer to object function f (x) prediction beyond flow
Customer flow actually uses the difference of flow with the user；If the customer flow of the prediction is more than, user is of that month actually to be made
It is more than the expense for exceeding flow with the expense of flow and the traffic policy for preparing to recommend, then with reference to the consumption of user
Capacity index parameter, so as to decide whether the traffic policy recommending the user；The consuming capacity index parameter is extremely
It is few to include following one kind：Sex, age, user identity, monthly use telephone expenses, average monthly income.
Here, if the customer flow of the prediction is more than the of that month actual use flow of user and described prepares what is recommended
The expense of traffic policy is more than the expense for exceeding flow, then the consuming capacity index parameter of user is needed to refer to, so as to certainly
Fixed the reason for whether traffic policy being recommended into the user, traffic policy here can be selected in practical application
Or the flow package of customization；The consumer group of different identity or occupation has different consuming capacity and consumption idea, for example, right
For student, if the of that month actual flow used may be not very willing to continue to buy beyond the flow in existing set meal
Flow package, but restraining uses flow in next month；For business people, if the of that month actual flow used is beyond existing
There is the flow in set meal, then will be considered that the flow of next month can use more, be interested in continuing with buying flow package.Certainly, institute is worked as
When stating the customer flows of object function f (x) predictions and being more than the of that month actual use flow of user, also can directly will be suitable flow
Strategy is recommended to user, without the consuming capacity index parameter with reference to user.
Further, methods described also includes：According to the historical forecast effect of the object function, to the object function
Optimize.
Here, the size of factor beta is estimated in the object function f (x) obtained in step 103 influences the object function f
(x) prediction effect, the β represent βjAnd βn.Therefore, in order to lift prediction effect, it is also necessary to being obtained in the step 103
The object function obtained optimizes, that is, optimizesWherein, y represents that user actually uses flow, g (x)
The customer flow demand of object function f (x) prediction is represented, u represents vectorial weight and u ≡ 1.
Assuming that λ represents the punishment parameter of the object function f (x), user's this month flow of the object function prediction needs
Ask as gT(x), user's traffic demand last month of prediction is gT-1(x), the object function is predicted in the prediction result of last month
User's collection of mistake is combined into Θ1, prediction correct user collection be combined into Θ2, the flow that the user of the prediction error refers to predict need to
The user for being less than or equal to predetermined threshold value with the difference of actual use flow is asked, the correct user of prediction refers to the flow of prediction
Demand and the difference of actual use flow are more than the user of predetermined threshold value, then can obtain vectorial weight u adaptive adjustment formula,
As shown in formula (1)：
Wherein, i ∈ Θ1Represent that user i belongs to the user in the prediction result mistake of last month；i∈Θ2Represent that user i belongs to
In the correct user of the prediction result of last month.
In summary, the optimization process for estimating factor beta is represented by：
Here, according to the optimization procedure expression for estimating factor beta, estimate factor beta to described and be constantly adjusted
With optimizing, so as to lift the prediction effect of the object function.
To realize the above method, the corresponding embodiment of the present invention additionally provides a kind of traffic policy recommendation apparatus, such as Fig. 2 institutes
Show, described device includes：Parameter structure module 21, object function establish module 22, recommending module 23；Wherein,
The parameter builds module 21, for the service condition parameter applied according to user to all kinds of APP, builds user couple
First affecting parameters of the service condition of all kinds of APP applications to customer flow demand；According to the basic use information parameter of user, structure
Build second affecting parameters of the basic use information of user to customer flow demand；
The object function establishes module 22, for according to first affecting parameters, the second affecting parameters, establishing prediction
The object function of customer flow demand；
The recommending module 23, for predicting user's request flow according to the object function, recommend flow plan to user
Slightly.
Wherein, the parameter structure module 21, is specifically used for：The service condition first applied according to user to all kinds of APP is joined
Number, obtains the preference index that user applies to all kinds of APP, and the preference index is included between preference weight, this month and last month partially
Good weight amplitude of variation；The service condition that the preference index applied further according to user to all kinds of APP, user apply to all kinds of APP
Flow service condition parameter in parameter, build first of user to the service condition that all kinds of APP are applied to customer flow demand
Affecting parameters, the flow service condition parameter include monthly use and use changes in flow rate amplitude between flow, this month and last month.
Here, when the service condition parameter that the user applies to all kinds of APP includes monthly access times, monthly use
Long, monthly use changes width between access times amplitude of variation, this month and last month between flow, this month and last month using duration
Changes in flow rate amplitude is used between degree, of that month and last month；
Wherein, the user refers to user within continuous N number of moon to such to every class APP monthly access times applied
The average access times of APP applications, the monthly use duration that the user applies to every class APP refer to user in continuous N number of moon
Interior to use duration to the average of such APP applications, the monthly use flow that the user applies to every class APP refers to user even
Continue in N number of moon and flow, N >=2 are used to the average of such APP applications；
The APP applications can be divided into 30 classifications, including social activity, Online Video, short-sighted frequency, game, music, reading, hand
Machine browser, Domestic News, ecommerce, fresh electric business, bank, tourism trip, financing, map, the trip of city bus, love and marriage
Friend-making, image beautification, cloud disk, mobile phone safe, cuisines, physical culture, automobile, household house ornamentation, house property, medical treatment ＆ health, body beautification body-building, female
Property, mother and baby, education, life weather.
Here, in being applied according to user to above-mentioned 30 class APP when the monthly access times of every class APP applications, monthly use
Long, monthly use changes width between access times amplitude of variation, this month and last month between flow, this month and last month using duration
Using 6 indexs such as changes in flow rate amplitudes between degree, of that month and last month, calculate what user applied to every class APP based on Information Entropy
Preference weight amplitude of variation, is described in detail below between preference weight, this month and last month：
Remember pi,1The monthly access times applied for user i to such APP, pi,2Such APP is applied for user i monthly
Use duration, pi,3Monthly flow, p are used for what user i applied to such APPi,4This month applied for user i to such APP
The access times amplitude of variation between last month, pi,5Become between last month for user i this month applied to such APP and using duration
Change amplitude, pi,6For user i this month applied to such APP and use changes in flow rate amplitude between last month.
1) according to formulaCalculate entropy H of the user to every class APP each indexs appliedk, whereinM represents number of users；
Here, to make the data p of different indexsi,kHomogeneity, can be to the data p of all different indexsi,kAdvanced rower is accurate
Change is handled, so as to facilitate subsequent treatment.
2) according to formulaCalculate entropy weight w of the user to every class APP each indexs appliedk；
3) according to formulaCalculate the preference ω ' that user applies to every class APPj, j=1,2 ..., 30；
4) according to formulaThe preference ω ' applied to the user to every class APPjIt is normalized, obtains
Take the preference weight ω that family is applied to every class APPj；
5) according to formula △ ωj=ωj(T)-ωj(T-1)Calculate preference between this month and last month that user applies to every class APP
Weight amplitude of variation △ ωj, wherein, T represents of that month, and T-1 represents the of that month last month.
Here, the preference weight amplitude of variation △ ωjUser can be embodied between of that month and last month to such APP
Using the variation tendency of preference, △ ωjIt is bigger, then it represents that user is bigger using the change of preference to such APP, right
The influence of customer flow demand is also more notable；△ωjIt is smaller, then it represents that user gets over to such APP using the change of preference
Small, the influence to customer flow demand is also got over not notable.
Here, it is represented by according to above-mentioned condition, first affecting parameters to customer flow demandWherein, f (ωj,△ωj) represent the preference weight that user applied in this month to jth class APP
ωjAnd preference weight amplitude of variation △ ω between corresponding of that month and last monthjAdjustment weight parameter under collective effect；βjRepresent
Coefficient is estimated to jth class APP applications, every class APP actual use situations applied can be configured according to user；φ(xh)
Represent the flow service condition parameter x that user applies to jth class APPhInfluence subparameter to customer flow demand, the flow
Service condition parameter xhIncluding monthly changes in flow rate amplitude is used using between flow, this month and last month；H represents that the flow makes
With the number of parameters included in situation parameter.
Here, the basic use information parameter of the user comprises at least following a kind of：Sex, age, occupation, monthly use
Telephone expenses, the of that month duration of call, the quantity of the of that month contact person using in flow, of that month contact contacts list, contacts list
In each contact person each contact person in monthly average is using flow, contacts list service condition ginseng that all kinds of APP applys
Number.
Here, second affecting parameters to customer flow demand are represented byWherein, βnRepresent to institute
State the parameter x included in the basic use information parameter of usernCorresponding estimates coefficient, can be according to the basic use information of user
Actual conditions be configured；Represent the parameter x included in the basic use information parameter of the usernTo customer flow
The influence subparameter of demand；R represents the number of parameters included in the basic use information parameter of user.
Here, each contact person is every in monthly average is using flow, contacts list in the contacts list of the user
Individual contact person may influence the flow use to the user itself on the service condition parameter that all kinds of APP are applied, and therefore, need
Build second affecting parameters of the basic use information of user to customer flow demand.
The object function establishes module 22, is specifically used for：The use obtained in module 21 is built according to the parameter
Family uses letter substantially to the service condition that all kinds of APP are applied to the first affecting parameters of customer flow demand and the user
The second affecting parameters to customer flow demand are ceased, the object function f of prediction customer flow demand is established using SAO-SVR algorithms
(x) it is, as follows：
Wherein, b is constant term.
The recommending module 23, is specifically used for：It is pre- that the object function that module 22 established is established according to the object function
Customer flow demand is surveyed, if the customer flow of the prediction is more than the of that month actual use flow of user and prepares the stream recommended
The expense of amount strategy is less than the expense for exceeding flow, then the traffic policy is recommended into the user；It is described to exceed stream
Amount refers to that the customer flow of object function f (x) prediction actually uses the difference of flow with the user；If the prediction
Customer flow is more than the of that month actual use flow of user and the expense for the traffic policy for preparing to recommend exceeds flow more than described
Expense, then with reference to the consuming capacity index parameter of user, so as to decide whether the traffic policy recommending the user；
The consuming capacity index parameter comprises at least following a kind of：Sex, age, user identity, monthly use telephone expenses, monthly receipts
Enter.
Here, if the customer flow of the prediction is more than the of that month actual use flow of user and prepares the flow recommended
The expense of strategy is more than the expense for exceeding flow, then the consuming capacity index parameter of user is needed to refer to, so as to determine to be
The no the reason for traffic policy is recommended into the user, traffic policy here can be selected or fixed in practical application
The flow package of system；The consumer group of different identity or occupation has different consuming capacity and consumption idea, for example, for learning
For life, if the of that month actual flow used may be not very willing to continue to buy flow beyond the flow in existing set meal
Set meal, but restraining uses flow in next month；For business people, if the of that month actual flow used is beyond existing set
Flow in meal, then will be considered that the flow of next month can use more, be interested in continuing with buying flow package.Certainly, when the mesh
, also can be directly by suitable traffic policy when the customer flow of scalar functions f (x) predictions is more than user's of that month actual use flow
Recommend to user, without the consuming capacity index parameter with reference to user.
Further, described device also includes：Optimization module 24, for being imitated according to the historical forecast of the object function
Fruit, the object function is optimized.
Here, the object function establishes in the object function f (x) that module 22 is obtained the size influence for estimating factor beta
The prediction effect of the object function f (x), the β represent βjAnd βn.Therefore, in order to lift prediction effect, it is also necessary to described
Object function optimizes, that is, optimizesWherein, y represents that user actually uses flow, and g (x) is represented
The customer flow demand of object function f (x) prediction, u represent vectorial weight and u ≡ 1.
Assuming that λ represents the punishment parameter of the object function f (x), user's this month flow of the object function prediction needs
Ask as gT(x), user's traffic demand last month of prediction is gT-1(x), the object function is predicted in the prediction result of last month
User's collection of mistake is combined into Θ1, prediction correct user collection be combined into Θ2, the flow that the user of the prediction error refers to predict need to
The user for being less than or equal to predetermined threshold value with the difference of actual use flow is asked, the correct user of prediction refers to the flow of prediction
Demand and the difference of actual use flow are more than the user of predetermined threshold value, then can obtain vectorial weight u adaptive adjustment formula,
As shown in formula (1)：
Wherein, i ∈ Θ1Represent that user i belongs to the user in the prediction result mistake of last month；i∈Θ2Represent that user i belongs to
In the correct user of the prediction result of last month.
In summary, the optimization process for estimating factor beta is represented by：
Here, according to the optimization procedure expression for estimating factor beta, estimate factor beta to described and be constantly adjusted
With optimizing, so as to lift the prediction effect of the object function.
In actual applications, the parameter structure module 21, object function establish module 22, recommending module 23, optimization mould
Block 24 can be by the central processing unit (CPU) positioned at terminal, microprocessor (MPU), digital signal processor (DSP) or scene
Programmable gate array (FPGA) etc. is realized.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.It is all
All any modification, equivalent and improvement made within the spirit and scope of the present invention etc., it is all contained in the protection model of the present invention
Within enclosing.
Claims (10)
1. a kind of traffic policy recommends method, it is characterised in that methods described includes：
The service condition parameter applied according to user to all kinds of APP, the service condition that structure user applies to all kinds of APP is to user
First affecting parameters of traffic demand；
According to the basic use information parameter of user, the structure basic use information of user influences to join on the second of customer flow demand
Number；
According to first affecting parameters, the second affecting parameters, the object function for predicting customer flow demand is established；
Customer flow demand is predicted according to the object function, recommends traffic policy to user.
2. according to the method for claim 1, it is characterised in that methods described also includes：
According to the historical forecast effect of the object function, the object function is optimized.
3. according to the method for claim 1, it is characterised in that
The service condition parameter was included between monthly access times, monthly use duration, monthly use flow, this month and last month
Used between access times amplitude of variation, this month and last month and changes in flow rate width is used between duration amplitude of variation, this month and last month
Degree；
The capital consumption information parameter comprises at least following a kind of：It is monthly to be flowed using telephone expenses, the of that month duration of call, of that month use
In amount, the quantity of contact person in of that month contact contacts list, contacts list each contact person when monthly average using flow,
The service condition parameter that each contact person applies to all kinds of APP in contacts list.
4. according to the method described in any one of claims 1 to 3, it is characterised in that described according to user all kinds of APP to be applied
Service condition parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand, bag
Include：
The service condition parameter applied according to user to all kinds of APP, obtain the preference index that user applies to all kinds of APP；It is described
Preference index includes preference weight amplitude of variation between preference weight, this month and last month；
Flow in the service condition parameter that the preference index applied according to user to all kinds of APP, user apply to all kinds of APP makes
With situation parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；The stream
Amount service condition parameter includes monthly use and uses changes in flow rate amplitude between flow, this month and last month.
5. according to the method for claim 1, it is characterised in that it is described to recommend traffic policy to user, including：
If the customer flow of the prediction is more than the of that month actual use flow of user and prepares the expense of traffic policy recommended
Less than the expense beyond flow, then the traffic policy is recommended into the user；It is described refer to beyond flow it is described pre-
The difference of the customer flow of survey and the actual use flow.
6. a kind of traffic policy recommendation apparatus, it is characterised in that described device includes：Parameter structure module, object function are established
Module, recommending module；Wherein,
The parameter builds module, and for the service condition parameter applied according to user to all kinds of APP, structure user is to all kinds of
First affecting parameters of the service condition of APP applications to customer flow demand；According to the basic use information parameter of user, structure is used
Second affecting parameters of the basic use information in family to customer flow demand；
The object function establishes module, for according to first affecting parameters, the second affecting parameters, establishing prediction user's stream
The object function of amount demand；
The recommending module, for predicting user's request flow according to the object function, recommend traffic policy to user.
7. device according to claim 6, it is characterised in that described device also includes：Optimization module, for according to
The historical forecast effect of object function, is optimized to the object function.
8. device according to claim 6, it is characterised in that
The service condition parameter was included between monthly access times, monthly use duration, monthly use flow, this month and last month
Used between access times amplitude of variation, this month and last month and changes in flow rate width is used between duration amplitude of variation, this month and last month
Degree；
The capital consumption information parameter comprises at least following a kind of：It is monthly to be flowed using telephone expenses, the of that month duration of call, of that month use
In amount, the quantity of contact person in of that month contact contacts list, contacts list each contact person when monthly average using flow,
The service condition parameter that each contact person applies to all kinds of APP in contacts list.
9. according to the device described in any one of claim 6 to 8, it is characterised in that the parameter builds module, is specifically used for：
The service condition parameter applied according to user to all kinds of APP, obtain the preference index that user applies to all kinds of APP；It is described
Preference index includes preference weight amplitude of variation between preference weight, this month and last month；
Flow in the service condition parameter that the preference index applied according to user to all kinds of APP, user apply to all kinds of APP makes
With situation parameter, first affecting parameters of the structure user to the service condition that all kinds of APP are applied to customer flow demand；The stream
Amount service condition parameter includes monthly use and uses changes in flow rate amplitude between flow, this month and last month.
10. device according to claim 6, it is characterised in that the recommending module, be specifically used for：
If the customer flow of the prediction is more than the of that month actual use flow of user and prepares the expense of traffic policy recommended
Less than the expense beyond flow, then the traffic policy is recommended into the user；It is described refer to beyond flow it is described pre-
The difference of the customer flow of survey and the actual use flow.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610654128.9A CN107729708A (en) | 2016-08-10 | 2016-08-10 | A kind of traffic policy recommends method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610654128.9A CN107729708A (en) | 2016-08-10 | 2016-08-10 | A kind of traffic policy recommends method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107729708A true CN107729708A (en) | 2018-02-23 |
Family
ID=61199502
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610654128.9A Pending CN107729708A (en) | 2016-08-10 | 2016-08-10 | A kind of traffic policy recommends method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107729708A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109284437A (en) * | 2018-08-01 | 2019-01-29 | 广东奥博信息产业股份有限公司 | A kind of adaptive feedback adjustment methods of weight and device of information push |
| CN110557339A (en) * | 2018-05-30 | 2019-12-10 | 阿里巴巴集团控股有限公司 | flow planning method and device, computer equipment and storage medium |
| CN115017400A (en) * | 2021-11-30 | 2022-09-06 | 荣耀终端有限公司 | Application APP recommendation method and electronic equipment |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103607691A (en) * | 2013-11-26 | 2014-02-26 | 中国联合网络通信集团有限公司 | Flow package recommendation method and device |
| CN104349367A (en) * | 2014-11-12 | 2015-02-11 | 深圳市中兴移动通信有限公司 | Mobile terminal and predicting reminding method and device of flow consumption of mobile terminal |
-
2016
- 2016-08-10 CN CN201610654128.9A patent/CN107729708A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103607691A (en) * | 2013-11-26 | 2014-02-26 | 中国联合网络通信集团有限公司 | Flow package recommendation method and device |
| CN104349367A (en) * | 2014-11-12 | 2015-02-11 | 深圳市中兴移动通信有限公司 | Mobile terminal and predicting reminding method and device of flow consumption of mobile terminal |
Non-Patent Citations (3)
| Title |
|---|
| 夏颖: "移动互联网用户价值的评价与分析方法", 《中国优秀硕士学位论文全文数据库_信息科技辑》 * |
| 朱顺应等: "《交通流参数及交通事件动态预测方法》", 31 May 2008 * |
| 秦洋: "大数据发展趋势下的中国电信运营商电子商务营销模式分析", 《中国优秀硕士学位论文全文数据库_经济与管理科学辑》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110557339A (en) * | 2018-05-30 | 2019-12-10 | 阿里巴巴集团控股有限公司 | flow planning method and device, computer equipment and storage medium |
| CN109284437A (en) * | 2018-08-01 | 2019-01-29 | 广东奥博信息产业股份有限公司 | A kind of adaptive feedback adjustment methods of weight and device of information push |
| CN109284437B (en) * | 2018-08-01 | 2020-12-08 | 广东奥博信息产业股份有限公司 | Weight adaptive feedback adjustment method and device for information push |
| CN115017400A (en) * | 2021-11-30 | 2022-09-06 | 荣耀终端有限公司 | Application APP recommendation method and electronic equipment |
| CN115017400B (en) * | 2021-11-30 | 2023-05-26 | 荣耀终端有限公司 | Application APP recommendation method and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jiao et al. | Profit maximization mechanism and data management for data analytics services | |
| CN105677831B (en) | Method and device for determining recommended merchants | |
| CN106097043B (en) | The processing method and server of a kind of credit data | |
| CN112633962B (en) | Service recommendation method and device, computer equipment and storage medium | |
| CN103425677B (en) | Keyword classification model determines method, keyword classification method and device | |
| CN104573304A (en) | User property state assessment method based on information entropy and cluster grouping | |
| CN107885886A (en) | To the method, apparatus and server of information recommendation sort result | |
| CN107071193A (en) | The method and apparatus of interactive answering system accessing user | |
| CN111797320B (en) | Data processing method, device, equipment and storage medium | |
| CN112417294A (en) | Intelligent business recommendation method based on neural network mining model | |
| CN107729708A (en) | A kind of traffic policy recommends method and device | |
| CN115994226B (en) | Clustering model training system and method based on federal learning | |
| CN111949887A (en) | Item recommendation method and device and computer-readable storage medium | |
| CN106846082A (en) | Tourism cold start-up consumer products commending system and method based on hardware information | |
| CN106991577A (en) | A kind of method and device for determining targeted customer | |
| CN116452304B (en) | Cross-domain green consumption scene integration and preferential recommendation method | |
| CN111626767B (en) | Resource data issuing method, device and equipment | |
| CN112966189A (en) | Fund product recommendation system | |
| CN107577736A (en) | A kind of file recommendation method and system based on BP neural network | |
| CN114417174B (en) | Content recommendation method, device, equipment and computer storage medium | |
| CN111078997B (en) | Information recommendation method and device | |
| CN110825974B (en) | Recommendation system content ordering method and device | |
| CN111861679A (en) | Commodity recommendation method based on artificial intelligence | |
| CN111695084A (en) | Model generation method, credit score generation method, device, equipment and storage medium | |
| CN107463853A (en) | The method and system of audient's label analysis |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180223 |
|
| RJ01 | Rejection of invention patent application after publication |