CN109583949B - User switching prediction method and system - Google Patents
User switching prediction method and system Download PDFInfo
- Publication number
- CN109583949B CN109583949B CN201811400416.7A CN201811400416A CN109583949B CN 109583949 B CN109583949 B CN 109583949B CN 201811400416 A CN201811400416 A CN 201811400416A CN 109583949 B CN109583949 B CN 109583949B
- Authority
- CN
- China
- Prior art keywords
- user
- information
- quantum
- quantum particle
- iteration
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 59
- 239000002245 particle Substances 0.000 claims abstract description 190
- 238000003066 decision tree Methods 0.000 claims abstract description 37
- 238000012549 training Methods 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 27
- 238000005457 optimization Methods 0.000 claims abstract description 23
- 239000002131 composite material Substances 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 33
- 239000000126 substance Substances 0.000 claims description 18
- 238000004458 analytical method Methods 0.000 claims description 16
- 230000006399 behavior Effects 0.000 claims description 10
- 230000002354 daily effect Effects 0.000 claims description 8
- 230000003203 everyday effect Effects 0.000 claims description 6
- 238000004590 computer program Methods 0.000 claims description 5
- 230000010365 information processing Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 6
- 239000002096 quantum dot Substances 0.000 description 6
- 230000000694 effects Effects 0.000 description 4
- 230000035772 mutation Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- OAICVXFJPJFONN-UHFFFAOYSA-N Phosphorus Chemical compound [P] OAICVXFJPJFONN-UHFFFAOYSA-N 0.000 description 2
- 238000013459 approach Methods 0.000 description 2
- 238000004891 communication Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G06Q50/40—
Abstract
The invention discloses a user change prediction method, a system and a device and a computer readable storage medium, and relates to the technical field of information processing. The method comprises the steps of coding the acquired user information, performing iterative optimization on the user information coding by adopting a composite parameter quantum particle swarm optimization algorithm, determining a user characteristic label as the input of a decision tree training model, training the user characteristic label according to the decision tree training model, determining an optimal objective function value, and outputting a user change prediction result. The method not only can process high-dimensional data, but also can improve the accuracy of the prediction result of the user change machine.
Description
Technical Field
The invention relates to the technical field of information processing, in particular to a method, a system and a device for predicting a user machine based on a composite parameter quantum particle swarm decision tree and a computer readable storage medium.
Background
At present, an intelligent terminal becomes an indispensable auxiliary tool for social contact, consumption, entertainment, information acquisition and the like in people's lives, with the rapid development of mobile communication technology and the intelligent terminal, the updating of the use scene, the brand, the model and the like of the mobile terminal is rapid, operators, equipment merchants and the like often need to perform terminal-related marketing activities, and in order to improve the accuracy of the switch prediction of telecommunication users, various telecommunication data sources need to be fully applied as much as possible.
The user switching prediction method in the prior art is mainly based on terminal shipment volume, user questionnaire survey, identification information sent by terminal startup, switching prediction by using the existing decision tree algorithm and other modes, and based on the terminal shipment volume mode, an analysis result is emphasized on the shipment volume and market occupation ratio of a mobile equipment manufacturer, and the condition that a user actually starts a new mobile equipment cannot be accurately analyzed and accurate switching position information cannot be obtained; based on the mode of user questionnaire survey, the analysis timeliness is poor due to the long period, and the application value of the analysis result is limited; based on the way of identification information sent by terminal startup, the strategy of user switching cannot be analyzed within a long time scale, and the accuracy of the analysis result obtained by the above way in the prior art is low; and the conventional decision tree algorithm is easy to over-fit when used for carrying out the machine change prediction, so that the actual prediction effect is not high, the actual prediction effect is not suitable for processing high-dimensional data, and when the number of attributes is too large, the decision tree effect is poor.
Disclosure of Invention
The embodiment of the invention provides a user change machine prediction method which can be applied to high-dimensional data and can improve the accuracy of a user change machine prediction result.
In order to achieve the above object, an embodiment of the present invention provides a user change machine prediction method, applied to a user change machine prediction apparatus, where the method includes:
acquiring user information, wherein the user information comprises: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information;
encoding the user information;
optimizing the codes according to a composite parameter quantum particle swarm optimization algorithm to obtain a user characteristic label;
and training the user characteristic label according to the decision tree training model to obtain a user change prediction result.
Correspondingly, the embodiment of the invention also provides a user change machine prediction system, which comprises an information acquisition unit, a coding unit and an analysis unit;
the information acquisition unit is configured to acquire user information, where the user information includes: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information;
the encoding unit is used for encoding the user information;
the analysis unit is used for optimizing the codes according to a composite parameter quantum particle swarm optimization algorithm to obtain a user characteristic label; and
and training the user characteristic label according to the decision tree training model to obtain a user change prediction result.
The invention adopts a composite parameter quantum particle swarm optimization algorithm to carry out iterative optimization on user information codes, obtains the input of a decision tree training model, namely a user characteristic label, trains the user characteristic label according to the decision tree training model, determines an optimal objective function value and outputs a user change prediction result. The method not only can process high-dimensional data, but also can improve the accuracy of the prediction result of the user change machine.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
FIG. 1 is a flow chart of a method for predicting a user's change of machine in accordance with an embodiment of the present invention;
FIG. 2 is a flow chart of a decision tree model algorithm according to an embodiment of the present invention;
FIG. 3 is a flow chart of a composite parameter quantum-behaved particle swarm optimization algorithm according to an embodiment of the invention;
FIG. 4 is a schematic diagram of a user switch machine prediction system according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a user change machine prediction apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, 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 invention.
An embodiment of the present invention provides a method for predicting a user changing machine, which is a flowchart of the method for predicting a user changing machine according to the present invention, as shown in fig. 1, where the method is applied to a device for predicting a user changing machine, and the method specifically includes:
s101, obtaining user information, wherein the user information comprises: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information.
The user information may be obtained from a user, the user information may also be obtained from a network side, or the user information may also be obtained from a third-party information platform, which is not limited in the embodiments of the present invention.
The terminal parameter information may include, but is not limited to, a terminal brand, a terminal model, a terminal price, a time to market, and an operating system; the terminal behavior information may include, but is not limited to, a terminal usage duration and whether the terminal is a contract machine; the user information may include, but is not limited to, user age, user gender, and user rating; the location information may include, but is not limited to, a user's day regular premises and a user's night regular premises; the consumption information may include, but is not limited to, average monthly consumption of the user, average monthly usage traffic of the user, average monthly talk time of the user, and user package type; the user preference information may include, but is not limited to, user common applications, user daily active time, number of tracking area lists TA List that the user passes through each day, and user attention category; the user browsing information may include, but is not limited to, the number and duration of times the user browses the mobile phone sales platform. The user information category can be adjusted according to specific situations. The embodiment of the present invention is not limited thereto.
And S102, encoding the user information.
Specifically, the discrete user information is subjected to one-hot coding by adopting an enumeration value method; and carrying out interval division on the continuous user information by adopting an equal frequency binning method, and coding each interval according to the interval division.
Wherein the discrete user information comprises: the method comprises the following steps of (1) terminal brand, terminal model, operating system, user gender, user grade, user daytime frequent site, user nighttime frequent site, user package type, user common application program, user daily active time and user attention category; the continuous user information includes: the method comprises the steps of terminal price, time to market, user age, average monthly consumption of a user, average monthly usage flow of the user, average monthly call duration of the user, the number of tracking area lists TA List passed by the user every day, and the times and duration of browsing a mobile phone sales platform by the user.
For example: the gender of the user is coded, male can be coded as 0, and female can be coded as 1; the terminal brand is coded, so that common TOP20 brands can be coded according to 1-20, and other brands are coded to be 0; the method comprises the steps of coding the daytime/nighttime frequent residence of a user, and coding different regions of different cities; the terminal price is encoded, and can be divided into intervals: (0, 2000], (2000, 4000], (4000, 6000], and (6000, + ∞)), and the above sections are encoded as 1, 2, 3, and 4, respectively.
S103, optimizing the codes according to a composite parameter quantum particle swarm optimization algorithm to obtain a user characteristic label.
And S104, training the user characteristic labels according to the decision tree training model to obtain a user change prediction result.
The process is specifically as shown in fig. 2, and is a flowchart of a decision tree model algorithm according to an embodiment of the present invention, and includes:
s1041, determining the input of the decision tree according to the user feature label;
in particular, according toDetermining a user characteristic label; wherein the content of the first and second substances,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]Are uniformly distributed with the random numbers in between,is the speed of the t iteration of the d bit of the ith quantum particle; wherein the content of the first and second substances,
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle; wherein
Wherein, a1And a2The constant is used for identifying the influence degree of the individual optimal position, the overall optimal position and the initialized position of the quantum particle on the speed of the quantum particle in the iteration.
Specifically, if the position of the ith bit of the ith quantum particle in the tth iterationIf 1, the corresponding user characteristic label is used as the input of the decision tree, and if the ith quantum isPosition of the tth iteration of the d bit of the particleIf 0, the user profile tag is discarded.
S1042, training the input of the decision tree according to the decision tree training model, if the iteration times is larger than a preset threshold, determining an optimal objective function value, and outputting a user change prediction result.
Further, as shown in fig. 3, it is a flowchart of the composite parameter quantum particle swarm optimization algorithm according to the embodiment of the present invention, that is, step S103 may include:
s1031, initializing the position, the speed and the individual optimal position of the quantum particles according to a composite parameter quantum particle swarm optimization algorithm; wherein the position x of the initialization quantum particlei0 or 1, initializing the velocity v of the quantum particlei=(αi1 αi2… αil)=(vi1 vi2 … vil) Initializing the individual optimal position p of the quantum particlei=xi(ii) a Wherein i refers to the ith quantum particle, and l refers to the dimension of the quantum particle, and the dimension is used for identifying the user information.
Wherein x isi=(xi1,xi2,…,xil),1≤i≤P;pi=(pi1 pi2 … pil)=(xi1,xi2,…,xil) Wherein P represents the scale of the quantum particle population, i.e., the number of quantum particles in the quantum particle population.
Specifically, for the conventional discrete particle swarm algorithm, the bits of the quantum particles can be represented as a pair of complex numbers (α, β), where α and β respectively represent probability ranges of different codes obtained after encoding the user information, where α is greater than or equal to 0 and less than or equal to 1, and | α |2+|β|21, the velocity of the quantum particle can therefore be expressed as:
vi=(αi1 αi2 … αil)=(vi1 vi2 … vil)。
the invention introduces quantityThe concept of the sub-rotating gate is to update the velocity of the quantum particle by using a quantum rotating angle theta, and the idea of the rotation is as follows: in the iterative process, the particles move to a larger direction of the objective function by rotating the angle theta, and if the position of the quantum particle is 1, which makes the objective function obtain a larger value, the | alpha | Y is2Should reduce |. beta |. non-conducting phosphor2Should be increased if | αi+1|2<|αi|2,|βi+1|2>|βi|2Then the quantum rotating gate is represented as:
where j represents the jth qubit of the particle, the velocity vij of the jth qubit of the ith particle can be expressed as:
it can also be expressed as:
wherein, a1And a2The constant is used for identifying the influence degree of the individual optimal position, the overall optimal position and the initialized position of the quantum particle on the speed of the quantum particle in the iteration.
Furthermore, in order to prevent the local optimal solution from being trapped, mutation probability c is introduced,
wherein mu is an attenuation factor, 0< mu <1 is satisfied, T is the maximum iteration number, and T represents the T-th iteration. The mutation probability c decreases as the number of iterations increases and approaches 0.
At this time, if the current position, the individual optimal position and the global optimal position of a certain bit of the quantum particle are the same and the random quantity r is satisfied<c, velocity v of j-th quantum bit of i-th particleijCan be expressed as:
where r is [0,1 ] randomly generated per iteration]And N is a quantum not gate. The formula can be simplified as:
thus, the speed of the tth iteration of the d bit of the ith quantum particle can be expressed as:
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle.
S1032, calculating an objective function of the quantum particles to obtain the overall optimal position p of the particle swarmg。
Wherein the overall optimal position p of the particle populationgRepresenting the whole particle swarm along with the movement of the whole particle swarm in a multidimensional spaceAnd calculating the objective function value of each quantum particle, and memorizing the optimal position reached by the particle with the optimal objective function value in all the particle swarms each time of iteration.
And S1033, updating the position and the speed of the quantum particle.
In particular, according to the formulaUpdating the velocity of the quantum particles, wherein,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle.
According to the formulaUpdating the position of the quantum particle, wherein,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]Are uniformly distributed with the random numbers in between,is the speed of the t-th iteration of the d-th bit of the ith quantum particle.
S1034, calculating an objective function of the quantum particles according to the updated positions and speeds of the quantum particles, and obtaining the individual optimal positions, the overall optimal positions and the average positions of the quantum particles.
In the process of multi-dimensional space motion, each particle memorizes the position of the maximum value of the objective function of each particle, and the individual optimal position of the quantum particle is used for representing the optimal position obtained by the particle.
The overall optimal position p of the particle swarmgAnd expressing the optimal positions reached by the particles which obtain the optimal objective function values in all the memorized particle swarms along with the calculation of the objective function values of each quantum particle and each iteration in the process of the motion of the whole particle swarms in the multidimensional space.
s1035, determining a user characteristic label according to the individual optimal position, the overall optimal position and the average position of the quantum particle.
Wherein the target function f (x) of the quantum particle is (TP + TN)/(TP + TN + FN + FP); wherein, TP represents that the prediction result is the user changing the machine, and the user actually changes the machine; FP indicates that the prediction result is that the user changes the machine, and the actual user does not change the machine; TN represents that the predicted user is changed and the actual user is not changed; FN indicates that the predicted user did not change the machine, the actual user changed the machine.
If the iteration times do not reach the preset maximum iteration times, repeating the steps S1033 and S1034; and if the iteration times reach the preset maximum iteration times, terminating the algorithm.
The invention adopts a composite parameter quantum particle swarm optimization algorithm to carry out iterative optimization on user information codes, obtains the input of a decision tree training model, namely a user characteristic label, trains the user characteristic label according to the decision tree training model, determines an optimal objective function value and outputs a user change prediction result. The method not only can process high-dimensional data, but also can improve the accuracy of the prediction result of the user change machine.
The embodiment of the present invention further provides a user change machine prediction system, as shown in fig. 4, which is a schematic structural diagram of the user change machine prediction system according to the embodiment of the present invention, where the user change machine prediction system 10 includes an information obtaining unit 110, a coding unit 120, and an analysis unit 130. Fig. 4 is a schematic diagram, and does not limit other modules of the user change machine prediction system and the structural relationship of each module. Wherein the content of the first and second substances,
an information obtaining unit 110, configured to obtain user information, where the user information includes: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information.
The user information may be obtained from a user, the user information may also be obtained from a network side, or the user information may also be obtained from a third-party information platform, which is not limited in the embodiments of the present invention.
The terminal parameter information may include, but is not limited to, a terminal brand, a terminal model, a terminal price, a time to market, and an operating system; the terminal behavior information may include, but is not limited to, a terminal usage duration and whether the terminal is a contract machine; the user information may include, but is not limited to, user age, user gender, and user rating; the location information may include, but is not limited to, a user's day regular premises and a user's night regular premises; the consumption information may include, but is not limited to, average monthly consumption of the user, average monthly usage traffic of the user, average monthly talk time of the user, and user package type; the user preference information may include, but is not limited to, the user's common applications, the user's daily active time, the number of tracking areas list, talests, the user passes through each day, and the user's category of interest; the user browsing information may include, but is not limited to, the number and duration of times the user browses the mobile phone sales platform. The user information category can be adjusted according to specific situations. The embodiment of the present invention is not limited thereto.
An encoding unit 120, configured to encode the user information.
Specifically, the discrete user information is subjected to one-hot coding by adopting an enumeration value method; and carrying out interval division on the continuous user information by adopting an equal frequency binning method, and coding each interval according to the interval division.
Wherein the discrete user information comprises: the method comprises the following steps of (1) terminal brand, terminal model, operating system, user gender, user grade, user daytime frequent site, user nighttime frequent site, user package type, user common application program, user daily active time and user attention category; the continuous user information includes: the method comprises the steps of terminal price, time to market, user age, average monthly consumption of a user, average monthly usage flow of the user, average monthly call duration of the user, the number of tracking area lists TAList passed by the user every day, and the times and duration of browsing a mobile phone sales platform by the user.
For example: the gender of the user is coded, male can be coded as 0, and female can be coded as 1; the terminal brand is coded, so that common TOP20 brands can be coded according to 1-20, and other brands are coded to be 0; the method comprises the steps of coding the daytime/nighttime frequent residence of a user, and coding different regions of different cities; the terminal price is encoded, and can be divided into intervals: (0, 2000], (2000, 4000], (4000, 6000], and (6000, + ∞)), and the above sections are encoded as 1, 2, 3, and 4, respectively.
The analysis unit 130 is configured to optimize the code according to a composite parameter quantum particle swarm optimization algorithm to obtain a user feature tag; and
and training the user characteristic label according to the decision tree training model to obtain a user change prediction result.
Specifically, the specific process of optimizing the code according to the composite parameter quantum particle swarm optimization algorithm to obtain the user feature label includes:
the analysis unit 130 initializes the position, speed and individual optimal position of the quantum particles according to the composite parameter quantum particle swarm optimization algorithm; wherein the position x of the initialization quantum particlei0 or 1, initializing the velocity v of the quantum particlei=(αi1 αi2 … αil)=(vi1 vi2 … vil) Initializing individual populations of quantum particlesPreferred position pi=xi(ii) a Wherein i refers to the ith quantum particle, and l refers to the dimension of the quantum particle, and the dimension is used for identifying the user information.
Wherein x isi=(xi1,xi2,…,xil),1≤i≤P;pi=(pi1 pi2 … pil)=(xi1,xi2,…,xil) Wherein P represents the scale of the quantum particle population, i.e., the number of quantum particles in the quantum particle population.
Specifically, for the conventional discrete particle swarm algorithm, the bits of the quantum particles can be represented as a pair of complex numbers (α, β), where α and β respectively represent probability ranges of different codes obtained after encoding the user information, where α is greater than or equal to 0 and less than or equal to 1, and | α |2+|β|21, the velocity of the quantum particle can therefore be expressed as:
vi=(αi1 αi2 … αil)=(vi1 vi2 … vil)。
the invention introduces the concept of quantum revolving door, updates the speed of quantum particles by using a quantum rotation angle theta, and the idea of the rotation is as follows: in the iterative process, the particles move to a larger direction of the objective function by rotating the angle theta, and if the position of the quantum particle is 1, which makes the objective function obtain a larger value, the | alpha | Y is2Should reduce |. beta |. non-conducting phosphor2Should be increased if | αi+1|2<|αi|2,|βi+1|2>|βi|2Then the quantum rotating gate is represented as:
where j represents the jth qubit of the particle, the velocity vij of the jth qubit of the ith particle can be expressed as:
it can also be expressed as:
wherein, a1And a2The constant is used for identifying the influence degree of the individual optimal position, the overall optimal position and the initialized position of the quantum particle on the speed of the quantum particle in the iteration.
Furthermore, in order to prevent the local optimal solution from being trapped, mutation probability c is introduced,
wherein mu is an attenuation factor, 0< mu <1 is satisfied, T is the maximum iteration number, and T represents the T-th iteration. The mutation probability c decreases as the number of iterations increases and approaches 0.
At this time, if the current position, the individual optimal position and the global optimal position of a certain bit of the quantum particle are the same and the random quantity r is satisfied<c, velocity v of j-th quantum bit of i-th particleijCan be expressed as:
where r is [0,1 ] randomly generated per iteration]And N is a quantum not gate. The formula can be simplified as:
thus, the speed of the tth iteration of the d bit of the ith quantum particle can be expressed as:
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle.
The analysis unit 130 calculates an objective function of the quantum particles to obtain a global optimal position p of the particle populationg。
Wherein the overall optimal position p of the particle populationgAnd expressing the optimal positions reached by the particles which obtain the optimal objective function values in all the memorized particle swarms along with the calculation of the objective function values of each quantum particle and each iteration in the process of the motion of the whole particle swarms in the multidimensional space.
The analysis unit 130 updates the position and velocity of the quantum particles.
In particular, according to the formulaUpdating the velocity of the quantum particles, wherein,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle.
According to the formulaUpdating the position of the quantum particle, wherein,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]Are uniformly distributed with the random numbers in between,is the speed of the t-th iteration of the d-th bit of the ith quantum particle.
The analysis unit 130 calculates an objective function of the quantum particle according to the updated position and velocity of the quantum particle, and obtains an individual optimal position, a total optimal position, and an average position of the quantum particle.
In the process of multi-dimensional space motion, each particle memorizes the position of the maximum value of the objective function of each particle, and the individual optimal position of the quantum particle is used for representing the optimal position obtained by the particle.
The overall optimal position p of the particle swarmgAnd expressing the optimal positions reached by the particles which obtain the optimal objective function values in all the memorized particle swarms along with the calculation of the objective function values of each quantum particle and each iteration in the process of the motion of the whole particle swarms in the multidimensional space.
the analysis unit 130 determines a user feature label according to the individual optimal position, the overall optimal position and the average position of the quantum particle;
wherein the target function f (x) of the quantum particle is (TP + TN)/(TP + TN + FN + FP); wherein, TP represents that the prediction result is the user changing the machine, and the user actually changes the machine; FP indicates that the prediction result is that the user changes the machine, and the actual user does not change the machine; TN represents that the predicted user is changed and the actual user is not changed; FN indicates that the predicted user did not change the machine, the actual user changed the machine.
If the iteration times do not reach the preset maximum iteration times, repeating the steps S1033 and S1034; and if the iteration times reach the preset maximum iteration times, terminating the algorithm.
Further, the training the user feature label according to the decision tree training model to obtain a user change prediction result includes:
the analysis unit 130 determines the input of the decision tree according to the user feature tag;
in particular, according toDetermining a user characteristic label; wherein the content of the first and second substances,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]Are uniformly distributed with the random numbers in between,is the speed of the t iteration of the d bit of the ith quantum particle; wherein the content of the first and second substances,
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the t-1 st overlap of the d bit of the ith quantum particleThe individual optimum position of the generation,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle; wherein
Wherein, a1And a2The constant is used for identifying the influence degree of the individual optimal position, the overall optimal position and the initialized position of the quantum particle on the speed of the quantum particle in the iteration.
Specifically, if the position of the ith bit of the ith quantum particle in the tth iterationIf the number is 1, the corresponding user feature label is used as the input of the decision tree, and if the position of the kth iteration of the d bit of the ith quantum particle is the position of the t iteration of the ith bitIf 0, the user profile tag is discarded.
The analysis unit 130 trains the input of the decision tree according to the decision tree training model, determines an optimal objective function value if the iteration number is greater than a preset threshold, and outputs a user switch prediction result.
The invention adopts a composite parameter quantum particle swarm optimization algorithm to carry out iterative optimization on user information codes, obtains the input of a decision tree training model, namely a user characteristic label, trains the user characteristic label according to the decision tree training model, determines an optimal objective function value and outputs a user change prediction result. The method not only can process high-dimensional data, but also can improve the accuracy of the prediction result of the user change machine.
The embodiment of the present invention further provides a user equipment change predicting apparatus, which is, as shown in fig. 5, a schematic structural diagram of the user equipment change predicting apparatus according to the embodiment of the present invention. The user change prediction apparatus 20 includes a memory 210 and a processor 220. Fig. 5 is a schematic diagram, and does not limit other modules of the user changing machine prediction apparatus and the structural relationship of each module. Wherein the content of the first and second substances,
a memory 210 for storing a computer program.
A processor 220 for executing the computer program to implement the user change prediction method.
In the several embodiments provided in the present application, it should be understood that the disclosed method and system may be implemented in other ways. For example, the above-described system embodiments are merely illustrative, and for example, the division of the functional blocks is only one logical division, and other divisions may be realized in practice, for example, multiple units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.
In addition, functional modules in the embodiments of the present invention may be integrated into one processing unit, or each unit may be physically included alone, or two or more units may be integrated into one unit. The integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.
The integrated unit implemented in the form of a software functional unit may be stored in a computer readable storage medium. The software functional unit is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute some steps of the methods according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and all the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.
Claims (10)
1. A method for predicting a user change machine, the method comprising:
acquiring user information, wherein the user information comprises: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information;
encoding the user information;
optimizing the codes according to a composite parameter quantum particle swarm optimization algorithm to obtain a user characteristic label;
training the user characteristic labels according to a decision tree training model to obtain a user change prediction result;
training the user characteristic label according to the decision tree training model to obtain a user change prediction result, wherein the training comprises the following steps:
determining input of the decision tree according to the user feature tag;
training the input of the decision tree according to a decision tree training model, if the iteration times are greater than a preset threshold, determining an optimal objective function value, and outputting a user change prediction result;
the optimizing the code according to the composite parameter quantum particle swarm optimization algorithm to obtain the user characteristic label comprises the following steps:
quantum particles according to composite parametersA group optimization algorithm is used for initializing the position, the speed and the individual optimal position of the quantum particles; wherein the position x of the initialization quantum particlei0 or 1, initializing the velocity v of the quantum particlei=(αi1 αi2…αil)=(vi1vi2…vil) Initializing the individual optimal position p of the quantum particlei=xi(ii) a Wherein i refers to the ith quantum particle, and l refers to the dimension of the quantum particle, and the dimension is used for identifying the user information;
the complex numbers (alpha, beta) are expressed as the bit of the quantum particle, wherein alpha and beta respectively express the probability amplitude of different codes obtained after the user information is coded, wherein alpha is more than or equal to 0 and less than or equal to 1, and | alpha is less than or equal to zero2+|β|2=1;
Calculating an objective function of the quantum particles to obtain an overall optimal position p of the particle swarmg;
Updating the position and the speed of the quantum particle;
calculating an objective function of the quantum particles according to the updated positions and speeds of the quantum particles to obtain the individual optimal positions, the overall optimal positions and the average positions of the quantum particles;
determining a user characteristic label according to the individual optimal position, the overall optimal position and the average position of the quantum particles;
wherein the target function f (x) of the quantum particle is (TP + TN)/(TP + TN + FN + FP); wherein, TP represents that the prediction result is the user changing the machine, and the user actually changes the machine; FP indicates that the prediction result is that the user changes the machine, and the actual user does not change the machine; TN represents the predicted user not changing the machine, and the actual user not changing the machine; FN indicates that the predicted user did not change the machine, the actual user changed the machine.
2. The method of claim 1, wherein the terminal parameter information comprises: a terminal brand, a terminal model, a terminal price, time to market and an operating system;
the terminal behavior information includes: the terminal use duration and whether the terminal is a contract machine;
the user information includes: user age, user gender, and user rating;
the location information includes: the user is normally located at day and night;
the consumption information includes: average monthly consumption of the user, average monthly usage flow of the user, average monthly call duration of the user and package type of the user;
the user preference information includes: the method comprises the following steps that a user commonly uses an application program, the daily active time of the user, the number of tracking area lists TA List passed by the user every day and the attention category of the user;
the user browsing information includes: and the times and duration of browsing the mobile phone sales platform by the user.
3. The method of claim 2, wherein the encoding the user information comprises:
carrying out one-hot coding on the discrete user information by adopting an enumeration value method; carrying out interval division on continuous user information by adopting an equal frequency binning method, and coding each interval according to the interval division;
wherein the discrete user information comprises: the method comprises the following steps of (1) terminal brand, terminal model, operating system, user gender, user grade, user daytime frequent site, user nighttime frequent site, user package type, user common application program, user daily active time and user attention category; the continuous user information includes: the method comprises the steps of terminal price, time to market, user age, average monthly consumption of a user, average monthly usage flow of the user, average monthly call duration of the user, the number of tracking area lists TA List passed by the user every day, and the times and duration of browsing a mobile phone sales platform by the user.
4. The user changing machine prediction method according to claim 1, wherein the determining a user feature label according to the individual optimal position, the overall optimal position, and the average position of the quantum particle comprises:
according toDetermining a user characteristic label; wherein the content of the first and second substances,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]Are uniformly distributed with the random numbers in between,is the speed of the t iteration of the d bit of the ith quantum particle; wherein the content of the first and second substances,
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle, c is the variation probability,wherein mu is an attenuation factor satisfying 0<μ<1, T is the maximum iteration number, and T represents the T-th iteration;
wherein, a1And a2Is a constant for identifying individual optimal positions, overall optimal positions and initialization of quantum particlesThe influence degree of the position of the quantum particle on the speed of the quantum particle in the iteration is shown, and the average position z is [ z ]1,z2,...,zM]Wherein, in the step (A),
5. a user change machine prediction system, the system comprising:
an information obtaining unit configured to obtain user information, the user information including: terminal parameter information, terminal behavior information, user information, location information, consumption information, user preference information, and user browsing information;
an encoding unit for encoding the user information;
the analysis unit is used for optimizing the codes according to a composite parameter quantum particle swarm optimization algorithm to obtain a user characteristic label; and
training the user characteristic labels according to a decision tree training model to obtain a user change prediction result;
training the user characteristic label according to the decision tree training model to obtain a user change prediction result, wherein the training comprises the following steps:
determining input of the decision tree according to the user feature tag;
training the input of the decision tree according to a decision tree training model, if the iteration times are greater than a preset threshold, determining an optimal objective function value, and outputting a user change prediction result;
the optimizing the code according to the composite parameter quantum particle swarm optimization algorithm to obtain the user characteristic label comprises the following steps:
initializing the position, the speed and the individual optimal position of the quantum particles according to a composite parameter quantum particle swarm optimization algorithm; wherein the position x of the initialization quantum particlei0 or 1, initializing the velocity v of the quantum particlei=(αi1 αi2…αil)=(vi1vi2…vil) Initializing individual optimal positions of quantum particlespi=xi(ii) a Wherein i refers to the ith quantum particle, and l refers to the dimension of the quantum particle, and the dimension is used for identifying the user information; the complex numbers (alpha, beta) are expressed as the bit of the quantum particle, wherein alpha and beta respectively express the probability amplitude of different codes obtained after the user information is coded, wherein alpha is more than or equal to 0 and less than or equal to 1, and | alpha is less than or equal to zero2+|β|2=1;
Calculating an objective function of the quantum particles to obtain an overall optimal position p of the particle swarmg;
Updating the position and the speed of the quantum particle;
calculating an objective function of the quantum particles according to the updated positions and speeds of the quantum particles to obtain the individual optimal positions, the overall optimal positions and the average positions of the quantum particles;
determining a user characteristic label according to the individual optimal position, the overall optimal position and the average position of the quantum particles;
wherein the target function f (x) of the quantum particle is (TP + TN)/(TP + TN + FN + FP); wherein, TP represents that the prediction result is the user changing the machine, and the user actually changes the machine; FP indicates that the prediction result is that the user changes the machine, and the actual user does not change the machine; TN represents the predicted user not changing the machine, and the actual user not changing the machine; FN indicates that the predicted user did not change the machine, the actual user changed the machine.
6. The system of claim 5, wherein the terminal parameter information comprises: a terminal brand, a terminal model, a terminal price, time to market and an operating system;
the terminal behavior information includes: the terminal use duration and whether the terminal is a contract machine;
the user information includes: user age, user gender, and user rating;
the location information includes: the user is normally located at day and night;
the consumption information includes: average monthly consumption of the user, average monthly usage flow of the user, average monthly call duration of the user and package type of the user;
the user preference information includes: the method comprises the following steps that a user commonly uses an application program, the daily active time of the user, the number of tracking area lists TA List passed by the user every day and the attention category of the user;
the user browsing information includes: and the times and duration of browsing the mobile phone sales platform by the user.
7. The user switch machine prediction system of claim 6, wherein the encoding the user information comprises:
carrying out one-hot coding on the discrete user information by adopting an enumeration value method; carrying out interval division on continuous user information by adopting an equal frequency binning method, and coding each interval according to the interval division;
wherein the discrete user information comprises: the method comprises the following steps of (1) terminal brand, terminal model, operating system, user gender, user grade, user daytime frequent site, user nighttime frequent site, user package type, user common application program, user daily active time and user attention category; the continuous user information includes: the method comprises the steps of terminal price, time to market, user age, average monthly consumption of a user, average monthly usage flow of the user, average monthly call duration of the user, the number of tracking area lists TA List passed by the user every day, and the times and duration of browsing a mobile phone sales platform by the user.
8. The user machine change prediction system of claim 5, wherein the determining a user feature label based on the individual optimal position, the overall optimal position, and the average position of the quantum particle comprises:
according toDetermining a user characteristic label; wherein the content of the first and second substances,is the position of the t iteration of the d bit of the ith quantum particle, and r is [0,1 ] randomly generated in each iteration]BetweenThe random numbers are distributed evenly and the random numbers are distributed evenly,is the speed of the t iteration of the d bit of the ith quantum particle; wherein the content of the first and second substances,
wherein the content of the first and second substances,is the rotation angle of the tth iteration of the d bit of the ith quantum particle;is the individual optimal position of the t-1 st iteration of the d bit of the ith quantum particle,is the overall optimal position of the t-1 st iteration of the d bit of the ith quantum particle, c is the variation probability,wherein mu is an attenuation factor satisfying 0<μ<1, T is the maximum iteration number, and T represents the T-th iteration;
wherein, a1And a2Is a constant for identifying the influence degree of the individual optimal position, the overall optimal position and the initialized position of the quantum particle on the speed of the quantum particle in the iteration, and the average position z is [ z ═1,z2,...,zM]Wherein, in the step (A),
9. a user change machine prediction apparatus, the apparatus comprising:
a memory for storing a computer program;
a processor for executing the computer program to implement the user change machine prediction method of any one of claims 1-4.
10. A computer-readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the user switch prediction method of any one of claims 1-4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811400416.7A CN109583949B (en) | 2018-11-22 | 2018-11-22 | User switching prediction method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811400416.7A CN109583949B (en) | 2018-11-22 | 2018-11-22 | User switching prediction method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109583949A CN109583949A (en) | 2019-04-05 |
CN109583949B true CN109583949B (en) | 2021-03-12 |
Family
ID=65923466
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811400416.7A Active CN109583949B (en) | 2018-11-22 | 2018-11-22 | User switching prediction method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109583949B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110175170B (en) * | 2019-04-19 | 2023-11-21 | 平安科技(深圳)有限公司 | Data optimization processing method, device, computer equipment and storage medium |
CN110991711A (en) * | 2019-11-19 | 2020-04-10 | 浙江鸿程计算机系统有限公司 | Multi-factor perception terminal switching prediction method based on deep neural network |
CN111047425B (en) * | 2019-11-25 | 2023-10-24 | 中国联合网络通信集团有限公司 | Behavior prediction method and device |
CN113538025B (en) * | 2020-04-14 | 2024-03-22 | 中国移动通信集团浙江有限公司 | Replacement prediction method and device for terminal equipment |
CN111913940B (en) * | 2020-06-20 | 2024-04-26 | 武汉海云健康科技股份有限公司 | Temperature membership tag prediction method and device, electronic equipment and storage medium |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104794195B (en) * | 2015-04-17 | 2018-04-06 | 南京大学 | A kind of data digging method found for the potential renewal user of telecommunications |
CN105678596B (en) * | 2016-03-23 | 2019-12-06 | 中国联合网络通信集团有限公司 | replacement prediction method and replacement prediction system for mobile terminal |
CN107783998A (en) * | 2016-08-26 | 2018-03-09 | 华为技术有限公司 | The method and device of a kind of data processing |
CN107016460A (en) * | 2017-03-27 | 2017-08-04 | 中国联合网络通信集团有限公司广西壮族自治区分公司 | User changes planes Forecasting Methodology and device |
US11030674B2 (en) * | 2017-04-14 | 2021-06-08 | International Business Machines Corporation | Cognitive order processing by predicting resalable returns |
CN107657267B (en) * | 2017-08-11 | 2021-11-09 | 百度在线网络技术(北京)有限公司 | Product potential user mining method and device |
CN107590569A (en) * | 2017-09-25 | 2018-01-16 | 山东浪潮云服务信息科技有限公司 | A kind of data predication method and device |
CN107798379B (en) * | 2017-11-23 | 2020-07-07 | 东北大学 | Method for improving quantum particle swarm optimization algorithm and application based on improved algorithm |
CN108256907A (en) * | 2018-01-09 | 2018-07-06 | 北京腾云天下科技有限公司 | A kind of construction method and computing device of customer grouping model |
CN108647364B (en) * | 2018-05-21 | 2021-10-29 | 广东工业大学 | Prediction recommendation method based on mobile terminal application data |
CN108846695A (en) * | 2018-06-07 | 2018-11-20 | 中国联合网络通信集团有限公司 | The prediction technique and device of terminal replacement cycle |
-
2018
- 2018-11-22 CN CN201811400416.7A patent/CN109583949B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN109583949A (en) | 2019-04-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109583949B (en) | User switching prediction method and system | |
CN111177579B (en) | Application method of integrated diversity enhanced ultra-deep factorization machine model | |
CN112100489B (en) | Object recommendation method, device and computer storage medium | |
CN112528164B (en) | User collaborative filtering recall method and device | |
CN111259222A (en) | Article recommendation method, system, electronic device and storage medium | |
CN111931053A (en) | Item pushing method and device based on clustering and matrix decomposition | |
CN113763093A (en) | User portrait-based item recommendation method and device | |
CN106909560A (en) | Point of interest sort method | |
CN112380449B (en) | Information recommendation method, model training method and related device | |
CN112053188B (en) | Internet advertisement recommendation method based on hybrid deep neural network model | |
CN114511387A (en) | Product recommendation method and device, electronic equipment and storage medium | |
CN110717788A (en) | Target user screening method and device | |
CN113688306A (en) | Recommendation strategy generation method and device based on reinforcement learning | |
CN117056591A (en) | Intelligent electric power payment channel recommendation method and system based on dynamic prediction | |
Jia et al. | Dynamic group recommendation algorithm based on member activity level | |
CN113409096B (en) | Target object identification method and device, computer equipment and storage medium | |
CN117194966A (en) | Training method and related device for object classification model | |
CN114357263A (en) | Method and device for processing multi-modal information of target object and storage medium | |
CN116739665A (en) | Information delivery method and device, electronic equipment and storage medium | |
CN113127755A (en) | Artificial intelligent virtual image information recommendation algorithm system and method | |
CN112308419A (en) | Data processing method, device, equipment and computer storage medium | |
CN111859117A (en) | Information recommendation method and device, electronic equipment and readable storage medium | |
CN112307319A (en) | Page generation method and device | |
Sable et al. | Reducing errors during stock value prediction Q-learning-based generic algorithm | |
CN116562359B (en) | CTR prediction model training method and device based on contrast learning and electronic equipment |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |