CN108648000A - Method and device, the electronic equipment that life cycle is assessed are retained to user - Google Patents
Method and device, the electronic equipment that life cycle is assessed are retained to user Download PDFInfo
- Publication number
- CN108648000A CN108648000A CN201810371862.3A CN201810371862A CN108648000A CN 108648000 A CN108648000 A CN 108648000A CN 201810371862 A CN201810371862 A CN 201810371862A CN 108648000 A CN108648000 A CN 108648000A
- Authority
- CN
- China
- Prior art keywords
- user
- complete
- clusters
- target user
- pattern
- 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.)
- Granted
Links
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
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0255—Targeted advertisements based on user history
-
- 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
- G06Q30/0241—Advertisements
- G06Q30/0251—Targeted advertisements
- G06Q30/0269—Targeted advertisements based on user profile or attribute
- G06Q30/0271—Personalized advertisement
-
- 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
- G06Q30/0241—Advertisements
- G06Q30/0272—Period of advertisement exposure
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Economics (AREA)
- Marketing (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Present invention is disclosed a kind of method and device assessed user's retention life cycle, electronic equipment, computer readable storage medium, the program includes:Obtain history alive data of the target user to technical routine;Duration is enlivened according to what history alive data was recorded, target user is divided to the identical a kind of user group for enlivening duration;The history alive data of all users in user group is clustered according to specified number of clusters, obtains the sub- lifetime value belonging to target user;Sub- lifetime value is matched one by one with the complete lifecycle pattern of specified number of clusters, determines the complete lifecycle pattern of target user.Program prediction through the invention obtains the complete lifecycle pattern of target user, it is more in line with the behavioural habits of target user, more meet the actual lifetime value of target user, it is possible thereby to solve the problems, such as the prior art can not Accurate Prediction subscriber lifecycle pattern, to effectively improve advertisement dispensing conversion ratio.
Description
Technical field
It is the present invention relates to field of computer technology, more particularly to a kind of to retain the method that life cycle is assessed to user
And device, electronic equipment, computer readable storage medium.
Background technology
In recent years, with the fast development of computer networking technology, internet has been widely used.User can lead to
Cross internet quickly and easily complete such as obtain information, shopping, payment, predetermined ticketing service it is various it is daily needed for, this to use
Family is increasingly strong to the dependence of internet.And for website, how in the retention life for knowing each user at the first time
Period will be that website policymaker is formulating product sale to obtain user in whole life cycle to the attention rate of client
It accurately, is timely supported with being provided in terms of publicity strategy.
It refers to that user is thoroughly taken off since with client opening relationships to client by internet to retain life cycle
Entire evolution from relationship.For example, as shown in Figure 1, the retention life-cycle processes of game user can be divided into 4 ranks
Section, examination object for appreciation phase form phase, stationary phase and are lost in the phase.By analyzing the conversion ratio in each stage, (number for completing conversion behavior accounts for
The ratio of the total number of clicks of promotion message) stage that high conversion can be found out, using the user in the stage as target user into
Row advertisement is launched, and the conversion ratio of advertisement can be increased substantially.
The process that the prior art is based primarily upon user's retention life cycle determines which stage user is in, then to height
The user in conversion ratio stage carries out advertisement dispensing, doing so this assumes that the rules of life cycle of different user is similar
, so as to be divided to the life-cycle processes of user using identical method.But actually distinct user has not
Same behavioural habits, the rule different as time change has of the input degree to client.And the prior art is due to can not be accurate
The life-cycle processes of different user really are predicted, and then can not targetedly carry out advertisement dispensing.
Invention content
In order to solve the problems, such as present in the relevant technologies can not Accurate Prediction subscriber lifecycle process, the present invention provides
It is a kind of to retain the method that life cycle is assessed to user.
On the one hand, the method that life cycle is assessed is retained to user the present invention provides a kind of, the method includes:
Obtain history alive data of the target user to technical routine;The history alive data includes enlivening duration;
Duration is enlivened according to what the history alive data was recorded, the target user is divided to and identical enlivens duration
A kind of user group;
The history alive data of all users in the user group is clustered according to specified number of clusters, obtains the target
Sub- lifetime value belonging to user;
The sub- lifetime value is matched one by one with the complete lifecycle pattern of specified number of clusters, described in determination
The complete lifecycle pattern of target user.
On the other hand, the device that life cycle is assessed, described device being retained to user the present invention also provides a kind of
Including:
Data acquisition module, for obtaining history alive data of the target user to technical routine;The history actively counts
According to including enlivening duration;
User's division module enlivens duration, by the target user for what is recorded according to the history alive data
It is divided to the identical a kind of user group for enlivening duration;
Pattern clustering module is carried out for the history alive data to all users in the user group according to specified number of clusters
Cluster, obtains the sub- lifetime value belonging to the target user;
Pattern Matching Module, for carrying out the complete lifecycle pattern of the sub- lifetime value and specified number of clusters
It matches one by one, determines the complete lifecycle pattern of the target user.
In addition, the present invention also provides a kind of electronic equipment, the electronic equipment includes:
Processor;
Memory for storing processor-executable instruction;
Wherein, the processor is configured as executing the above-mentioned method for assessing user's retention life cycle.
In addition, the present invention also provides a kind of computer readable storage medium, the computer-readable recording medium storage
There are computer program, the computer program that can retain what life cycle was assessed to user by processor execution completion is above-mentioned
Method.
The technical solution that the embodiment of the present invention provides can include the following benefits:
The present invention embodies the general character between user by excavating the complete lifecycle pattern of specified number of clusters.Then will
According to the sub- lifetime value that the history alive data of target user is generated, the complete lifecycle pattern with specified number of clusters
It is matched one by one, so as to obtain the prediction result of target user's complete lifecycle pattern.It predicts in this way
The complete lifecycle pattern of target user is obtained, the behavioural habits of target user are more in line with, more meets target user's reality
Lifetime value, and then the user of high conversion can be found out according to the lifetime value of user, realize the standard of advertisement
Really launch, overcome the prior art due to can not Accurate Prediction user lifetime value so that cause advertisement not launch accurately
The problem of.
It should be understood that above general description and following detailed description is merely exemplary, this can not be limited
Invention.
Description of the drawings
The drawings herein are incorporated into the specification and forms part of this specification, and shows the implementation for meeting the present invention
Example, and in specification together principle for explaining the present invention.
Fig. 1 is the life-cycle processes schematic diagram for the game user that the prior art provides;
Fig. 2 is the schematic diagram according to implementation environment according to the present invention;
Fig. 3 is a kind of block diagram of server shown according to an exemplary embodiment;
Fig. 4 is a kind of method flow assessed user's retention life cycle shown according to an exemplary embodiment
Figure;
Fig. 5 is the application scenarios schematic diagram provided in an embodiment of the present invention assessed user's retention life cycle;
Fig. 6 is the details flow chart of the step 410 of Fig. 4 corresponding embodiments;
Fig. 7 is the details flow chart of the step 450 of Fig. 4 corresponding embodiments;
Fig. 8 is the details flow chart of the step 470 of Fig. 4 corresponding embodiments;
Fig. 9 is provided by the invention another to user's retention life cycle progress on the basis of Fig. 8 corresponding embodiments
The method flow diagram of assessment;
Figure 10 is the state diagram of the complete lifecycle of a variety of game users;
Figure 11 is the cluster result comparison diagram of different cluster class number of clusters;
Figure 12 be on the basis of Fig. 9 corresponding embodiments it is provided by the invention it is another to user retain life cycle into
The method flow diagram of row assessment;
Figure 13 is the complete alive data cluster result comparison diagram of different life length;
Figure 14 is the relationship comparison diagram of 4 kinds of complete lifetime values and game type of game user
Figure 15 is the cluster result comparison diagram of more game users and single game user;
Figure 16 is the comparison diagram that conversion ratio is launched in advertisement;
Figure 17 is that a kind of method assessed user's retention life cycle shown according to an exemplary embodiment fills
The block diagram set;
Figure 18 is the details block diagram of data acquisition module in Figure 17 corresponding embodiments;
Figure 19 is the details block diagram of pattern clustering module in Figure 17 corresponding embodiments;
Figure 20 is the details block diagram of Pattern Matching Module in Figure 17 corresponding embodiments.
Specific implementation mode
Here will explanation be executed to exemplary embodiment in detail, the example is illustrated in the accompanying drawings.Following description is related to
When attached drawing, unless otherwise indicated, the same numbers in different drawings indicate the same or similar elements.Following exemplary embodiment
Described in embodiment do not represent and the consistent all embodiments of the present invention.On the contrary, they be only with it is such as appended
The example of the consistent device and method of some aspects being described in detail in claims, of the invention.
Fig. 2 is the implementation environment schematic diagram according to the present invention shown according to an exemplary embodiment.Involved by the present invention
And implementation environment include server 210.Be stored in the database of server 210 target user history alive data and
The complete lifecycle pattern of specified number of clusters provided by the invention retains Life Cycle to which server 210 may be used to user
The method that phase is assessed, the history alive data based on target user obtain the sub- lifetime value of target user, then
Sub- lifetime value is matched with the complete lifecycle pattern of specified number of clusters, obtains the complete life belonging to target user
Order cyclic pattern.
As needed, which can also include providing data, i.e. the complete life of history alive data, specified number of clusters
Order the source of the data such as cyclic pattern.Specifically, in this implementation environment, data source can be mobile terminal 230.Service
Device 210 can receive history alive data from mobile terminal 230, specify the complete lifecycle pattern of number of clusters can be by mobile whole
End 230 is directly sent to server 210, can also be the multiple sample of users provided according to mobile terminal 230 by server 210
Complete alive data generate.
It should be noted that provided by the invention retain user the method that life cycle is assessed, it is not limited to taking
It is engaged in disposing corresponding processing logic in device 210, can also be the processing logic being deployed in other machines.For example, having
The processing logic etc. of subscriber lifecycle model prediction is disposed in the terminal device of computing capability.
It is a kind of server architecture schematic diagram provided in an embodiment of the present invention referring to Fig. 3, Fig. 3.The server 300 can be because matching
It sets or performance is different and generate bigger difference, may include one or more central processing units (central
Processing units, CPU) 322 (for example, one or more processors) and memory 332, one or more
Store the storage medium 330 (such as one or more mass memory units) of application program 342 or data 344.Wherein, it deposits
Reservoir 332 and storage medium 330 can be of short duration storage or persistent storage.The program for being stored in storage medium 330 may include
One or more modules (diagram is not shown), each module may include to the series of instructions operation in server 300.
Further, central processing unit 322 could be provided as communicating with storage medium 330, and storage medium is executed on server 300
Series of instructions operation in 330.Server 300 can also include one or more power supplys 326, one or more
Wired or wireless network interface 350, one or more input/output interfaces 358, and/or, one or more operations
System 341, such as Windows ServerTM, Mac OSXTM, UnixTM,LinuxTM, FreeBSDTMEtc..Following Fig. 4, Fig. 6-
9, the server architecture shown in Fig. 3 can be based on by the step performed by server described in embodiment illustrated in fig. 12.
One of ordinary skill in the art will appreciate that realizing that all or part of step of following embodiments can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can be stored in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
Fig. 4 is a kind of method flow assessed user's retention life cycle shown according to an exemplary embodiment
Figure.This retains user the scope of application and executive agent for the method that life cycle is assessed, for example, this method is used for Fig. 2
The server 210 of shown implementation environment.As shown in figure 4, the prediction technique, can be executed by server 210, may include following
Step.
In step 410, history alive data of the target user to technical routine is obtained;The history alive data includes
Enliven duration.
Wherein, it refers to that user once participates in the project to the end from participation technical routine for the first time that user, which retains life cycle,
In the process, daily participation situation (such as participating in probability).Target user refers to carrying out retaining lifetime value assessment
User.Technical routine can be the application program that certain money is participated in for user, for example, Games Software app, social networking application class app,
Personal consumption class app, audio-visual broadcast message class app etc., to assess user using these application programs retention life cycle (including
Headquarters of the General Staff and number of days, daily participation probability etc.).As needed, technical routine can also be a certain movement of user's participation, Yong Hucan
With a certain work etc., to assess total number of days and the daily participation probability that user participates in a certain movement or work.
History alive data refers to before the lifetime value of prediction target user, and target user is active in this daily
The state of technical routine.Such as target user day logs in and specifies Games Software app, it is believed that same day alive data is 1, such as
Fruit day is not logged on record, it is believed that this day alive data is 0.The Games Software is used from target user's registration in first day
App starts, and until when obtaining history alive data, user daily will be used as the alive data of this Games Software app should
The history alive data of target user.Similarly, target user day takes part in a certain movement, it is believed that same day alive data is
1, it is not engaged within second day this movement and can consider that this day alive data is 0, and so on.Join within first day from target user
Start with this movement, until when obtaining history alive data, the daily alive data of user will be used as the target user to join
With the history alive data of this movement.
Wherein, it refers to that the target user recorded in history alive data enlivens number of days to technical routine to enliven duration,
I.e. target user is already engaged in the number of days of the technical routine.For example, target user's registration in first day logs in certain Games Software app and arrives
Total number of days when current acquisition history alive data.
Specifically, alive data of the target user to technical routine can also in real time be acquired by mobile terminal 230, and report
To server 210, server 210 can record the alive data reported in real time in the local database, and what is reported daily enlivens
Situation constitutes history alive data.When needing to predict the lifetime value of target user, server 210 can be from local number
According to acquisition target user in library to the history alive data of technical routine.
In step 430, duration is enlivened according to what the history alive data was recorded, the target user is divided to
The identical a kind of user group for enlivening duration;
Wherein, user group refers to by multiple one group of user constituted with the identical user for enlivening duration.It needs to illustrate
It is, it is above-mentioned to can consider the user with certain life cycle length, at this time with the target user for centainly enlivening duration
It is in life cycle, but without the complete life cycle data of formation, therefore similarity calculation or rail cannot be passed through
The methods of mark similarity is matched with the complete lifecycle pattern of specified number of clusters, judges the complete life belonging to target user
Order cyclic pattern.
The present invention by way of clustering and judging, according to target user it is current enliven duration, target user is divided to
With the identical a kind of user group for enlivening duration.In other words, the current duration that enlivens of all users is equal in the user group
's.Assuming that the duration that enlivens that target user records in history alive data is 5 weeks, then when by the target user with current active
Length is that 5 weeks users are divided into one group.
In step 450, the history alive data of all users in the user group is clustered according to specified number of clusters,
Obtain the sub- lifetime value belonging to the target user;
Wherein, the number of clusters of cluster indicates the number of lifetime value.Specified number of clusters can be by advance to great amount of samples
The complete alive data of user clusters, if generate three kinds of complete lifetime values that number of clusters is 3, the area between cluster
Indexing is high, and number distribution is more reasonable, then it can be 3 to specify number of clusters.Likewise, if when number of clusters is 4, the discrimination between cluster
Higher, number distribution is more reasonable, then it can be 4 to specify number of clusters.
According to above-mentioned specified number of clusters, the history alive data of all users in user group is gathered by clustering algorithm
Class.For example, the user of user group can be polymerized to by four classes by clustering algorithm, obtains the sub- life cycle mould of four class users
Formula, and then for the classification of target user's ownership, obtain the sub- lifetime value of target user.
It is target user from starting to contact the technical routine it should be noted that sub- lifetime value is incomplete
To the lifetime value at current time.For sub- lifetime value is with respect to complete lifecycle pattern.Such as it plays
The complete lifecycle pattern of user refers to certain a game, and a user finally no longer plays this game from coming into contact with
Whole process.Sub- lifetime value is a part for this whole process.User is to the input degree of game with the time
Variation has certain rule.It can be generally divided into examination object for appreciation phase, formation phase, stationary phase and be lost in the phase.And sub- lifetime value can
Phase and formation phase are played in the examination that front can be only included.
In step 470, the complete lifecycle pattern of the sub- lifetime value and specified number of clusters is carried out one by one
Matching, determines the complete lifecycle pattern of the target user.
It should be noted that the complete lifecycle pattern of specified number of clusters refers to being used great amount of samples by clustering algorithm
The complete alive data at family is clustered, the cluster result of obtained specified number of clusters.Complete alive data is exactly complete life
It is daily during the entire process of the project to being finally no longer participate in from starting to contact a certain project that cycle data refers to sample of users
Participation state.
Assuming that specified number of clusters is 4, that is to say, that by being clustered to great amount of samples data, generate 4 kinds of complete life
Cyclic pattern.The sub- lifetime value of target user is matched with each complete lifecycle pattern respectively, can be obtained
To with the most matched complete lifecycle pattern of sub- lifetime value.The matched complete lifecycle pattern is exactly according to mesh
The complete lifecycle cyclic pattern for the target user that the history alive data of mark user predicts.Based on the complete of the target user
Whole lifetime value can predict the target user later daily to the participation probability of technical routine.Also, by complete
Whole lifetime value is divided, and can also find out the period of high conversion, and then targetedly carry out advertisement dispensing.
The whole life-cycle processes that the prior art gives tacit consent to all users are similar, are then based on life shown in FIG. 1
Periodic process finds out the period of high conversion, advertisement dispensing is carried out to the user in the period, to which different degrees of raising advertisement is thrown
Put rate.But practical different user has different behavioural habits, to the input degree of game in different times with different
Variation, if the rules of life cycle for directly giving tacit consent to all users is all similar, will directly affect each subscriber lifecycle
The accuracy of pattern, and then can not accurately mark off the period progress advertisement dispensing of high conversion.
It should be noted that, although each game user has a different personal gaming life cycles, but these users
Between can have general character, therefore the present invention is by excavating several representative game user lifetime value (i.e. above-mentioned fingers
Determine the complete lifecycle pattern of number of clusters) embody the general character between these users.Then by the sub- life cycle of target user
Pattern is matched with above-mentioned complete lifecycle pattern, can predict the complete lifecycle pattern belonging to target user.
The complete lifecycle pattern of the target user predicted in this way, is more in line with the behavioural habits of target user.
Further, based on the above-mentioned complete lifecycle pattern for being more in line with user behavior custom, high conversion is marked off
The period of rate carries out advertisement dispensing, can utmostly improve ad conversion rates.
Fig. 5 is the application scenarios schematic diagram provided in an embodiment of the present invention assessed user's retention life cycle.Such as
Shown in Fig. 5, which includes data platform 510, media provision side's platform 520 (SSP), advertisement transaction platform 530 (ADX)
And want advertisement side's platform 540 (DSP).Want advertisement side's platform 540 services for advertiser, and advertiser can be needed by advertisement
The target audience of advertisement, dispensing region, advertisement bid etc. is arranged in the side's of asking platform 540.520 service advertisement position of media provision side's platform
Owning side, possess rich-media resource and customer flow media can on this platform the advertisement position of oneself, control it is wide
That accuses shows, setting complementary etc..Advertisement transaction platform 530 connects want advertisement side's platform 540 and media provision side's platform
520。
Wherein, data platform DMP, Data Management Platform) 510 can be want advertisement side's platform 540
User characteristic data is provided, consequently facilitating want advertisement side's platform 540 targetedly selects the target audience of advertisement.Further
, the method provided by the invention assessed user's retention life cycle may be used in data platform 510, analyzes a large amount of samples
The life cycle data of this user excavates the complete lifecycle pattern of specified number of clusters and corresponding transformation rule, in turn
According to the history alive data of target user, the complete lifecycle pattern of target user can be predicted, to effectively distinguish
The user of different conversion capabilities.
Want advertisement side's platform 540 be directed to different user conversion capability, the user that conversion capability can be selected high as
User is launched in advertisement.Advertisement transaction platform 530 can select suitable advertisement according to the bid etc. of want advertisement side's platform 540,
The advertisement is controlled from media provision side's platform 520 to above-mentioned advertisement dispensing user to show.Thus.It realizes to high conversion capability
User carries out advertisement dispensing, substantially increases the conversion ratio of advertisement dispensing, and advertiser is helped to maximize ROI (rate of return on investment),
Form the benign cycle that advertisement is launched.
Fig. 6 is a kind of method stream assessed user's retention life cycle shown according to another exemplary embodiment
Cheng Tu.As shown in fig. 6, the step 410 in Fig. 4 corresponding embodiments specifically includes:
In step 411, participation state recording of the target user to the technical routine is obtained;
Wherein, the state that the target user that state recording refers to record participates in the technical routine daily is participated in.Participation state
It is divided into participation and is not involved in.Assuming that target user participates in the state recording moment from the technical routine to current obtain is participated in for the first time,
It is 25 days total, then it can get in 25 days, whether target user participates in the state of the technical routine daily.
When user participates in the technical routine, can by affiliated 230 real-time report of mobile terminal of user to server 210, by
210 real-time update of server participates in state recording.When needing to carry out lifetime value prediction to target user, server
210 have the participation state recording that storage is obtained in own database.
In step 412, by carrying out sequential encoding to the participation state recording, the target user is generated to described
The history alive data of technical routine.
To participating in state recording to carry out sequential encoding referring to daily giving birth to the participation state of technical routine according to target user
At corresponding character.For example, participating in then generating number 1, it is not involved in, generates number 0.By suitable to participating in state recording progress
Sequence encodes, and can generate such as character string as 11101010110, and each number represents the participation state on the same day, is generated
Character string can be as target user to the history alive data of technical routine.
Fig. 7 is a kind of method stream assessed user's retention life cycle shown according to another exemplary embodiment
Cheng Tu.As shown in fig. 7, the step 450 in Fig. 4 corresponding embodiments specifically includes:
In step 451, the history alive data of all users in the user group is gathered by spectral clustering
Class generates the sub- lifetime value for specifying number of clusters;
Wherein, spectral clustering is established on the basis of spectral graph theory, and compared with traditional clustering algorithm, it has can be in office
The advantages of being clustered on the sample space for shape of anticipating and converging on globally optimal solution.The algorithm is first according to given data set definition
One is described as the affinity matrix to data point similarity, and the characteristic value and feature vector of calculating matrix.Then selection is closed
The different data point of suitable feature vector clusters.Assuming that by the spectral clustering to the complete alive data of great amount of samples user
When being clustered, it is 4 preferably to cluster number of clusters, then can be by above-mentioned spectral clustering to the history of all users in user group
Alive data is clustered, and obtains 4 classes cluster as a result, generating the sub- lifetime value of specified number of clusters.
In step 452, from the sub- lifetime value of the specified number of clusters, the son belonging to the target user is obtained
Lifetime value.
Assuming that the sub- lifetime value of 4 classes is generated, the classification belonged in cluster according to target user, you can obtain mesh
Mark the sub- lifetime value of user.
It should be noted that for some new users, such as just participated in certain game just two weeks, state recording is participated at this time
Data volume it is less, life cycle characteristic curve also unobvious can if being that these users divide sub- lifetime value by force
The prediction of mistake can be generated.According to the conversion ratio of existing lifetime value different times it is found that being opened when user is in game
When beginning to try the object for appreciation phase, the conversion ratio that advertisement is launched is very high, it is possible thereby to individually by the less user of the data volume for participating in state recording
It is divided into one kind, advertisement dispensing is individually carried out to this kind of user, the conversion ratio of advertisement can also be improved.
Fig. 8 is a kind of method stream assessed user's retention life cycle shown according to another exemplary embodiment
Cheng Tu.As shown in figure 8, the step 470 in Fig. 4 corresponding embodiments specifically includes:
In step 471, calculate the sub- lifetime value of the target user and each complete lifecycle pattern it
Between similarity;
It is participated in it should be noted that each complete lifecycle pattern may include user from last time is participated in for the first time
In the complete procedure of some project, daily participation probability.And sub- lifetime value may include user is participated in from for the first time
Current time participates in the probability of project daily.
Assuming that can be clustered to obtain by the complete alive data of all sample of users by spectral clustering, 4 kinds complete raw
Cyclic pattern is ordered, then can generate each complete lifecycle according to the daily participation probability of each complete lifecycle pattern
The feature vector a of pattern.According to participation probability daily in the sub- lifetime value of target user, feature vector b is generated, is passed through
The Euclidean distance or cosine similarity for calculating feature vector a and b, obtain sub- lifetime value and each complete lifecycle
Similarity between pattern.
In step 472, from the complete lifecycle pattern of specified number of clusters, filter out and the sub- lifetime value
The highest complete lifecycle pattern of similarity, obtains the complete lifecycle pattern of the target user.
Assuming that specified number of clusters is 4, according to the similarity between sub- lifetime value and each complete lifecycle pattern,
It can be filtered out highest complete with the similarity of the sub- lifetime value of target user from 4 kinds of complete lifetime values
Whole lifetime value.Using the selection result as the prediction result of target user's complete lifecycle pattern.
Fig. 9 is a kind of method stream assessed user's retention life cycle shown according to a further exemplary embodiment
Cheng Tu.As shown in figure 9, on the basis of Fig. 4 corresponding embodiments, can also include the following steps before step 470:
In step 901, the complete alive data of multiple sample of users is obtained;
Wherein, multiple sample of users may include participating in a variety of users of disparity items.For example with game item, more
A sample of users may include the user for participating in sport game, the user for participating in the game of racing class, participate in card cards game
User and participation role play the part of class user etc..Complete alive data refers to sample of users from participating in a certain project for the first time to the end
During once participating in the project, daily actual participation state.If participate in the day and can mark be, if being not involved in this
It can be marked, the character string for 1010001 this forms that thus complete alive data can be 110100 ..., each number
First day participation state is represented successively, participates within second day state, third day participates in state,,, participate in state within n-th day.
In step 902, according to the complete alive data of the multiple sample of users, filter out with isometric life cycle
And it is only active in the candidate user of unitem simultaneously;
It should be noted that in order to effectively analyze the whole life cycle state of user, then choosing has complete live
The sample of users for the data that jump.Wherein, according to the analysis of existing subscriber lifecycle it is found that the length of empty window phase is three weeks, also
It is to say, if a user is not engaged in technical routine in continuous three weeks, the life cycle for being considered as the user terminates.As a result, may be used
To obtain user from the alive data for beginning participating in the whole process once participated in the end.
Further, in order to exclude the data source isomerism that the life cycle of different length is brought, can select to have etc.
The user of long life cycle, that is to say, that enliven the identical user of number of days in complete alive data.In addition, simultaneously due to user
Influence of the number of entry of participation to analysis is unknown, it is possible thereby to only select while being active in the user of unitem.And then it can
From multiple sample of users, the user for meeting above two condition is filtered out, as candidate user.Specific data format is such as
Under:
User 1:(the 1st day use state, the 2nd day use state ..., n-th day use state)
User 2:(the 1st day use state, the 2nd day use state ..., n-th day use state)
…
User m:(the 1st day use state, the 2nd day use state ..., n-th day use state);
In step 903, the complete alive data of all candidate users is clustered, generates the complete of the specified number of clusters
Whole lifetime value.
Specifically, can divide all candidate users according to the game classification of participation, marks off a variety of game and use
Family, and the user of each type game according to the complete alive data of each candidate user, can be calculated in life cycle
Interior daily participation probability.The state diagram of the whole life cycle of a variety of game users of Figure 10.It should be noted that due to game
Type is more, and clearly to embody the lifetime value difference of different type game, Figure 10 schematically lists wherein 5 kinds
User's entirety life cycle state of different type game respectively includes role playing 1001, shooting-flight shooting 1002, chess
Board 1003, strategy-tower anti-1004, the game of the classifications such as racing 1005.Abscissa expression enlivens number of days, and ordinate indicates ginseng daily
With probability.It can be seen from the figure that it is 100% that first day participates in probability with last day, and interim, the game of each classification
Participation probability it is different, curve shown in Fig. 10 is clustered by spectral clustering, can generate number of clusters size conjunction
The curve of suitable a variety of complete lifecycle patterns.
Optionally, the complete alive data of all candidate users is clustered by spectral clustering, three can be generated
Kind or four kinds of complete lifetime values.
It is to be understood that number of clusters how much representatives generate complete lifecycle pattern type how much.Such as Figure 11 institutes
Show, is clustered to enlivening the complete alive data of candidate user that number of days is 60 days, when it is 2 to cluster number of clusters, the second cluster excavates
And it is insufficient, there is larger refinement space, because two classes can be further subdivided into when number of clusters is 3.When number of clusters be 5 and its with
On, most of cluster excessively refines, and number accounting is minimum, will appear the low problem of coverage rate in practical applications, so not
Properly.When number of clusters is 3 or 4, the discrimination between cluster is higher, and number distribution is more reasonable.Therefore, thicker if doing granularity
Analysis can choose number of clusters be equal to 3, general analysis can choose number of clusters be 4.
Further, on the basis of Fig. 9 corresponding embodiments, as shown in figure 12, after step 903, step 470 it
Before, the method provided by the invention assessed user's retention life cycle can also include the following steps:
In step 1201, obtains with different life length, is active in disparity items classification while being active in more
A project or the verification user for being only active in unitem simultaneously;
It should be noted that the correctness of the complete lifecycle pattern of specified number of clusters in order to verify above-mentioned generation, main
Will from different life length, be active in disparity items classification while being active in multiple projects while being active in single project
Etc. factors analyzed.As a result, verification user may include life cycle length (completely enlivening number of days) be 60 days, 90 days and
120 days three classes users.It may include being active in disparity items classification, such as be active in different game classification (bodies to verify user
Educate class game, card cards game, Role Playing Game and intelligence development class game etc.) a few class users.Verifying user may include
Multiple projects are active in simultaneously, such as simultaneously participate in the user of a variety of game.It can be active in single item simultaneously to verify user
Mesh, such as single game user.By control variate method, obtains have a kind of verification user of condition every time.
In step 1202, the complete alive data of the verification user is clustered according to the specified number of clusters, is obtained
Obtain the cluster result of the specified number of clusters;
With reference to above-described embodiment, the complete alive data for verifying user is carried out according to specified number of clusters using spectral clustering
Cluster.Such as specified number of clusters can be 4, cluster result in obtaining 4.
In step 1203, by the complete lifecycle pattern of the cluster result of the specified number of clusters and the specified number of clusters
It is compared, verifies the correctness of the complete lifecycle pattern.
Specifically, above-mentioned 4 kinds of cluster results and generated 4 kinds complete lifetime values are compared one by one, test
Demonstrate,prove the correctness of each complete lifecycle pattern.For example, can by calculate cluster result and complete lifecycle pattern it
Between similarity, similarity be more than threshold value when, it is believed that be verified, complete lifecycle pattern is correct.
Figure 13 is the complete alive data cluster result comparison diagram of different life length.It can be observed from fig. 13 that
When complete lifecycle length is 60 days, 90 days and 120 days, although complete lifecycle length is different, it is polymerized to the knot of 4 classes
Fruit is completely identical, and it is smaller to be indicated above influence of the different life length to complete lifecycle pattern.
Figure 14 is the relationship comparison diagram of 4 kinds of complete lifetime values and game type of game user.Due to practical trip
Type of playing is more, and as shown in figure 14, only listing 6 kinds of classifications, (strategy-tower is anti-, role playing-MMO, and chess and card, shooting-flight are penetrated
Hit, leisure-elimination, intelligence development) game schematically illustrated, by being to participating in above-mentioned game and complete lifecycle length
The 60 complete alive data of user clusters.As shown in figure 14, be copolymerized into 4 classes (label0, label1, label2,
Label3 number accounting analysis), and to every class cluster result is carried out, 6 kinds of game classifications are respectively shared in label0 cluster results
Number ratio is close, in label1 cluster results 6 kinds of game classifications respectively shared number ratio also close to label2 cluster results
In 6 kinds of game classifications respectively shared number ratio still relatively, the respective shared people of 6 kinds of game classifications in label3 cluster results
Number ratios also close to.It is possible thereby to find that number distribution of the different game classifications on same cluster result is much like, that is,
Say that different game classification discriminations in cluster are small.It can be considered that shadow of the project category participated in complete lifecycle pattern
Sound is small.
Figure 15 is the cluster result comparison diagram of more game users and single game user, as shown in figure 15, by single game
The life cycle data of player and more game players cluster, it is possible to find their cluster result is all similar, is equally said
Bright, influence of the factor to complete lifecycle pattern is little.
It to sum up, can to the analysis of experimental results of life cycle length, game classification, single game player and more game players
To know, the complete lifecycle pattern for the specified number of clusters that the method provided through the invention is generated is not affected by these factors,
With stronger generalization ability.As a result, no matter the life cycle length of target user is how many, which the game item of participation is
Kind, if it is more game players, by the way that the sub- lifetime value of target user is matched with complete lifecycle pattern,
Can Accurate Prediction go out the complete lifecycle pattern of target user.
As needed, advertisement can be launched by the user to each lifetime value, according to the feedback of these users
Data analyze the ad conversion rates of the user of each lifetime value.For launching first game advertisement, conversion ratio
Definition can be game excited user number/exposure number of users.
Figure 16 is the comparison diagram that conversion ratio is launched in advertisement.As can be seen from Figure 16, it is in user's (figure pilot scale object for appreciation of examination object for appreciation phase
Phase group) conversion ratio be relatively high, be consistent with existing conclusion.By using method provided by the invention, 4 kinds are excavated completely
The user of each lifetime value is divided into one group, is represented sequentially as lable-0, lable-1 by lifetime value,
Lable-2, lable-3.As shown in figure 16, the high conversion rate of the user group of label-1 in launch at random advertisement group (i.e. in figure with
Machine dispensing group) effect, and the conversion ratio of other three groups of users launches advertisement group less than random.This explanation is by predicting that target is used
Complete lifecycle pattern belonging to family can efficiently differentiate the user of high conversion and low-conversion, and then for high conversion
The target user of rate carries out game advertisement dispensing, and advertiser is helped to maximize ROI, forms the benign cycle that advertisement is launched.
It further, can also be further to each complete lifecycle if the feedback data of advertisement dispensing is more
Pattern is refined, and the process of each complete lifecycle pattern is divided.Such as by the complete Life Cycle of lable-0 groups
Phase mode division is at two stages, growth stage and stationary phase.By the complete lifecycle mode division of label-1 groups at two ranks
Section, is familiar with phase and active period etc..Such division, can more subtly find high conversion period, and when being converted to height
The user of phase carries out advertisement dispensing, improves ad conversion rates.
Following is apparatus of the present invention embodiment, can be used for executing being stayed to user for the above-mentioned execution of server 210 of the present invention
Deposit the embodiment of the method that life cycle is assessed.For undisclosed details in apparatus of the present invention embodiment, this hair is please referred to
It is bright that the embodiment of the method that life cycle is assessed is retained to user.
Figure 17 is a kind of device assessed user's retention life cycle shown according to an exemplary embodiment
Block diagram, this is retained the device that life cycle is assessed to user and can be used in the server 210 of implementation environment shown in Fig. 2,
The all or part of step for the method that life cycle is assessed is retained shown in execution Fig. 4, Fig. 6-9, Figure 12 is any to user
Suddenly.As shown in figure 17, which includes but not limited to:Data acquisition module 1710, user's division module 1730, pattern clustering mould
Block 1750, Pattern Matching Module 1770.
Data acquisition module 1710, for obtaining history alive data of the target user to technical routine;The history is lived
Jump data include enlivening duration;
User's division module 1730 enlivens duration, by the target for what is recorded according to the history alive data
User is divided to the identical a kind of user group for enlivening duration;
Pattern clustering module 1750, for the history alive data to all users in the user group according to specified number of clusters
It is clustered, obtains the sub- lifetime value belonging to the target user;
Pattern Matching Module 1770 is used for the complete lifecycle pattern of the sub- lifetime value and specified number of clusters
It is matched one by one, determines the complete lifecycle pattern of the target user.
The function of modules and the realization process of effect specifically refer to above-mentioned to user's retention Life Cycle in above-mentioned apparatus
The realization process of step is corresponded in the method that phase is assessed, details are not described herein.
Data acquisition module 1710 such as can be some physical arrangement input/output interface 358 in Fig. 3.
User's division module 1730, pattern clustering module 1750, Pattern Matching Module 1770 can also be function module, use
In the above-mentioned correspondence step retained to user in the method that life cycle is assessed of execution.It is appreciated that these modules can be with
By hardware, software, or a combination of both realize.When realizing in hardware, these modules may be embodied as one or more
A hardware module, such as one or more application-specific integrated circuits.When being realized with software mode, these modules may be embodied as
The one or more computer programs executed in one or more processors, such as depositing performed by the central processing unit 322 of Fig. 3
Store up the program in memory 332.
In a kind of exemplary embodiment, as shown in figure 18, the data acquisition module 1710 includes:
State acquiring unit 1711, for obtaining participation state recording of the target user to the technical routine;
State encoding unit 1712, for by carrying out sequential encoding to the participation state recording, generating the target
History alive data of the user to the technical routine.
In a kind of exemplary embodiment, as shown in figure 19, the pattern clustering module 1750 includes:
Subpattern cluster cell 1751, for active to the history of all users in the user group by spectral clustering
Data are clustered, and the sub- lifetime value for specifying number of clusters is generated;
Subpattern obtaining unit 1752, for from the sub- lifetime value of the specified number of clusters, obtaining the target
Sub- lifetime value belonging to user.
In a kind of exemplary embodiment, as shown in figure 20, the Pattern Matching Module 1770 includes:
Similarity calculated 1771, the sub- lifetime value for calculating the target user and each complete life
Similarity between cyclic pattern;
Screening unit 1772 is matched, it is raw with the son for from the complete lifecycle pattern of specified number of clusters, filtering out
The highest complete lifecycle pattern of cyclic pattern similarity is ordered, the complete lifecycle pattern of the target user is obtained.
Further, the above-mentioned device assessed user's retention life cycle can also include:
Sample acquisition module, the complete alive data for obtaining multiple sample of users;
User's screening module, for according to the complete alive data of the multiple sample of users, filter out with etc. it is long-living
Life period and the candidate user for being only active in unitem simultaneously;
User clustering module is clustered for the complete alive data to all candidate users, generates the specified cluster
Several complete lifecycle patterns.
Wherein, the user clustering module may include:
Spectral clustering unit, it is raw for being clustered to the complete alive data of all candidate users by spectral clustering
At three kinds or four kinds of complete lifetime values.
Further, the above-mentioned device assessed user's retention life cycle can also include:
User's acquisition module is verified, for obtaining with different life length, being active in disparity items classification, simultaneously
It is active in multiple projects or is only active in the verification user of unitem simultaneously;
Cluster result obtains module, is carried out according to the specified number of clusters for the complete alive data to the verification user
Cluster obtains the cluster result of the specified number of clusters;
Cluster result contrast module is used for the complete life of the cluster result of the specified number of clusters and the specified number of clusters
Cyclic pattern is compared, and the correctness of the complete lifecycle pattern is verified.
Optionally, the present invention also provides a kind of electronic equipment, which can be used for the clothes of implementation environment shown in Fig. 1
Be engaged in device 210, execute Fig. 4, Fig. 6-9, Figure 12 it is any shown in the whole of method that life cycle is assessed is retained to user
Or part steps.The electronic equipment includes:
Processor;
Memory for storing processor-executable instruction;
Wherein, the processor is configured as executing the above-mentioned method for assessing user's retention life cycle.
The processor of electronic equipment executes the concrete mode operated and retains life to user in related this in the embodiment
Detailed description is performed in the embodiment for the method that the life period is assessed, explanation will be not set forth in detail herein.
In the exemplary embodiment, a kind of storage medium is additionally provided, which is computer readable storage medium,
Such as can be the provisional and non-transitorycomputer readable storage medium for including instruction.The storage medium is stored with computer
Program, the computer program can be given birth to by that can be executed by the central processing unit 322 of server 300 with completing above-mentioned retained to user
The method that the life period is assessed.
It should be understood that the invention is not limited in the precision architectures for being described above and being shown in the accompanying drawings, and
And various modifications and change can be being executed without departing from the scope.The scope of the present invention is limited only by the attached claims.
Claims (15)
1. a kind of retaining user the method that life cycle is assessed, which is characterized in that the method includes:
Obtain history alive data of the target user to technical routine;The history alive data includes enlivening duration;
Duration is enlivened according to what the history alive data was recorded, the target user is divided to and identical enlivens the one of duration
Class user group;
The history alive data of all users in the user group is clustered according to specified number of clusters, obtains the target user
Affiliated sub- lifetime value;
The sub- lifetime value is matched one by one with the complete lifecycle pattern of specified number of clusters, determines the target
The complete lifecycle pattern of user.
2. according to the method described in claim 1, it is characterized in that, the acquisition target user is active to the history of technical routine
Data, including:
Obtain participation state recording of the target user to the technical routine;
By carrying out sequential encoding to the participation state recording, generates the target user and live to the history of the technical routine
Jump data.
3. according to the method described in claim 1, it is characterized in that, the history to all users in the user group is active
Data are clustered according to specified number of clusters, obtain the sub- lifetime value belonging to the target user, including:
The history alive data of all users in the user group is clustered by spectral clustering, generates and specifies number of clusters
Sub- lifetime value;
From the sub- lifetime value of the specified number of clusters, the sub- lifetime value belonging to the target user is obtained.
4. according to the method described in claim 1, it is characterized in that, described by the sub- lifetime value and specified number of clusters
Complete lifecycle pattern is matched one by one, determines the lifetime value of the target user, including:
Calculate the similarity between the sub- lifetime value of the target user and each complete lifecycle pattern;
From the complete lifecycle pattern of specified number of clusters, filter out highest complete with the sub- lifetime value similarity
Lifetime value obtains the complete lifecycle pattern of the target user.
5. according to the method described in claim 1, it is characterized in that, described by the sub- lifetime value and specified number of clusters
Complete lifecycle pattern is matched one by one, before the complete lifecycle pattern for determining the target user, the method
Further include:
Obtain the complete alive data of multiple sample of users;
According to the complete alive data of the multiple sample of users, filters out with isometric life cycle and be only active in list simultaneously
The candidate user of one project;
The complete alive data of all candidate users is clustered, the complete lifecycle pattern of the specified number of clusters is generated.
6. according to the method described in claim 5, it is characterized in that, the complete alive data of described pair of all candidate users carries out
Cluster generates the complete lifecycle pattern of the specified number of clusters, including:
The complete alive data of all candidate users is clustered by spectral clustering, generates three kinds or four kinds of complete life
Cyclic pattern.
7. according to the method described in claim 5, it is characterized in that, the complete alive data of described pair of all candidate users carries out
It clusters, after the complete lifecycle pattern for generating the specified number of clusters, the method further includes:
It obtains with different life length, be active in disparity items classification while being active in multiple projects or only living simultaneously
It jumps in the verification user of unitem;
The complete alive data of the verification user is clustered according to the specified number of clusters, obtains the poly- of the specified number of clusters
Class result;
The cluster result of the specified number of clusters and the complete lifecycle pattern of the specified number of clusters are compared, described in verification
The correctness of complete lifecycle pattern.
8. a kind of retaining user in the device that life cycle is assessed, described device includes:
Data acquisition module, for obtaining history alive data of the target user to technical routine;The history alive data packet
It includes and enlivens duration;
User's division module enlivens duration for what is recorded according to the history alive data, the target user is divided
To the identical a kind of user group for enlivening duration;
Pattern clustering module is gathered for the history alive data to all users in the user group according to specified number of clusters
Class obtains the sub- lifetime value belonging to the target user;
Pattern Matching Module, for carrying out one by one the complete lifecycle pattern of the sub- lifetime value and specified number of clusters
Matching, determines the complete lifecycle pattern of the target user.
9. device according to claim 8, which is characterized in that the data acquisition module includes:
State acquiring unit, for obtaining participation state recording of the target user to the technical routine;
State encoding unit, for by carrying out sequential encoding to the participation state recording, generating the target user to institute
State the history alive data of technical routine.
10. device according to claim 8, which is characterized in that the pattern clustering module includes:
Subpattern cluster cell, for being carried out to the history alive data of all users in the user group by spectral clustering
Cluster generates the sub- lifetime value for specifying number of clusters;
Subpattern obtaining unit, for from the sub- lifetime value of the specified number of clusters, obtaining belonging to the target user
Sub- lifetime value.
11. device according to claim 8, which is characterized in that the Pattern Matching Module includes:
Similarity calculated, the sub- lifetime value for calculating the target user and each complete lifecycle pattern
Between similarity;
Screening unit is matched, for from the complete lifecycle pattern of specified number of clusters, filtering out and the sub- life cycle mould
The highest complete lifecycle pattern of formula similarity, obtains the complete lifecycle pattern of the target user.
12. device according to claim 8, which is characterized in that described device further includes:
Sample acquisition module, the complete alive data for obtaining multiple sample of users;
User's screening module is filtered out for the complete alive data according to the multiple sample of users with isometric Life Cycle
Phase and the candidate user for being only active in unitem simultaneously;
User clustering module is clustered for the complete alive data to all candidate users, generates the specified number of clusters
Complete lifecycle pattern.
13. device according to claim 12, which is characterized in that the user clustering module includes:
Spectral clustering unit generates three for being clustered to the complete alive data of all candidate users by spectral clustering
Kind or four kinds of complete lifetime values.
14. a kind of electronic equipment, which is characterized in that the electronic equipment includes:
Processor;
Memory for storing processor-executable instruction;
Wherein, the processor be configured as perform claim require 1-7 any one described in user retain life cycle into
The method of row assessment.
15. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer journey
Sequence, the computer program can be executed by processor to be completed to retain Life Cycle to user described in claim 1-7 any one
The method that phase is assessed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810371862.3A CN108648000B (en) | 2018-04-24 | 2018-04-24 | Method and device for evaluating user retention life cycle and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810371862.3A CN108648000B (en) | 2018-04-24 | 2018-04-24 | Method and device for evaluating user retention life cycle and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108648000A true CN108648000A (en) | 2018-10-12 |
CN108648000B CN108648000B (en) | 2022-10-28 |
Family
ID=63747145
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810371862.3A Active CN108648000B (en) | 2018-04-24 | 2018-04-24 | Method and device for evaluating user retention life cycle and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108648000B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109754290A (en) * | 2019-01-15 | 2019-05-14 | 网易(杭州)网络有限公司 | A kind for the treatment of method and apparatus of game data |
CN110148011A (en) * | 2019-04-15 | 2019-08-20 | 平安科技(深圳)有限公司 | Active amount decline analysis method, device, equipment and medium based on big data |
CN111090706A (en) * | 2018-10-24 | 2020-05-01 | 北京嘀嘀无限科技发展有限公司 | Computer storage medium, and method and system for acquiring life cycle of driver platform |
CN111489003A (en) * | 2019-01-25 | 2020-08-04 | 华为技术有限公司 | Life cycle prediction method and device |
CN111833110A (en) * | 2020-07-23 | 2020-10-27 | 北京思特奇信息技术股份有限公司 | Customer life cycle positioning method and device, electronic equipment and storage medium |
CN112232862A (en) * | 2020-10-12 | 2021-01-15 | 上海移卓网络科技有限公司 | Method, device, equipment and storage medium for predicting advertisement business channel ROI |
CN112685674A (en) * | 2020-12-30 | 2021-04-20 | 百果园技术(新加坡)有限公司 | Feature evaluation method and device influencing user retention |
CN113626705A (en) * | 2021-08-10 | 2021-11-09 | 广州虎牙科技有限公司 | User retention analysis method and device, electronic equipment and storage medium |
CN114742569A (en) * | 2021-01-08 | 2022-07-12 | 广州视源电子科技股份有限公司 | User life stage prediction method and device, computer equipment and storage medium |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104759100A (en) * | 2015-03-23 | 2015-07-08 | 小米科技有限责任公司 | Prediction method and device of game life cycle and game operator server |
CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
US20160328226A1 (en) * | 2015-05-08 | 2016-11-10 | Desktop 365, LLC | Method and system for managing the end to end lifecycle of the virtualization environment for an appliance |
CN106408217A (en) * | 2016-11-10 | 2017-02-15 | 北京京东金融科技控股有限公司 | Product life cycle identification method and device |
CN107451832A (en) * | 2016-05-30 | 2017-12-08 | 北京京东尚科信息技术有限公司 | The method and apparatus of pushed information |
CN107784390A (en) * | 2017-10-19 | 2018-03-09 | 北京京东尚科信息技术有限公司 | Recognition methods, device, electronic equipment and the storage medium of subscriber lifecycle |
-
2018
- 2018-04-24 CN CN201810371862.3A patent/CN108648000B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104759100A (en) * | 2015-03-23 | 2015-07-08 | 小米科技有限责任公司 | Prediction method and device of game life cycle and game operator server |
US20160328226A1 (en) * | 2015-05-08 | 2016-11-10 | Desktop 365, LLC | Method and system for managing the end to end lifecycle of the virtualization environment for an appliance |
CN107451832A (en) * | 2016-05-30 | 2017-12-08 | 北京京东尚科信息技术有限公司 | The method and apparatus of pushed information |
CN106097114A (en) * | 2016-06-27 | 2016-11-09 | 江苏易乐网络科技有限公司 | A kind of game public sentiment monitoring and player's behavior analysis cloud service platform |
CN106408217A (en) * | 2016-11-10 | 2017-02-15 | 北京京东金融科技控股有限公司 | Product life cycle identification method and device |
CN107784390A (en) * | 2017-10-19 | 2018-03-09 | 北京京东尚科信息技术有限公司 | Recognition methods, device, electronic equipment and the storage medium of subscriber lifecycle |
Non-Patent Citations (2)
Title |
---|
李彪等: "新世纪以来新闻传播学研究的生命周期及学术权力地图(2001-2016)――基于科学知识图谱的分析", 《国际新闻界》 * |
赵国锋等: "移动社交网的生命周期评估模型研究", 《计算机学报》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111090706A (en) * | 2018-10-24 | 2020-05-01 | 北京嘀嘀无限科技发展有限公司 | Computer storage medium, and method and system for acquiring life cycle of driver platform |
CN109754290A (en) * | 2019-01-15 | 2019-05-14 | 网易(杭州)网络有限公司 | A kind for the treatment of method and apparatus of game data |
CN109754290B (en) * | 2019-01-15 | 2024-06-04 | 网易(杭州)网络有限公司 | Game data processing method and device |
CN111489003B (en) * | 2019-01-25 | 2023-04-18 | 华为技术有限公司 | Life cycle prediction method and device |
CN111489003A (en) * | 2019-01-25 | 2020-08-04 | 华为技术有限公司 | Life cycle prediction method and device |
CN110148011B (en) * | 2019-04-15 | 2024-03-22 | 平安科技(深圳)有限公司 | Method, device, equipment and medium for analyzing active amount drop based on big data |
CN110148011A (en) * | 2019-04-15 | 2019-08-20 | 平安科技(深圳)有限公司 | Active amount decline analysis method, device, equipment and medium based on big data |
CN111833110A (en) * | 2020-07-23 | 2020-10-27 | 北京思特奇信息技术股份有限公司 | Customer life cycle positioning method and device, electronic equipment and storage medium |
CN112232862A (en) * | 2020-10-12 | 2021-01-15 | 上海移卓网络科技有限公司 | Method, device, equipment and storage medium for predicting advertisement business channel ROI |
CN112685674A (en) * | 2020-12-30 | 2021-04-20 | 百果园技术(新加坡)有限公司 | Feature evaluation method and device influencing user retention |
CN114742569A (en) * | 2021-01-08 | 2022-07-12 | 广州视源电子科技股份有限公司 | User life stage prediction method and device, computer equipment and storage medium |
CN113626705A (en) * | 2021-08-10 | 2021-11-09 | 广州虎牙科技有限公司 | User retention analysis method and device, electronic equipment and storage medium |
CN113626705B (en) * | 2021-08-10 | 2023-10-20 | 广州虎牙科技有限公司 | User retention analysis method, device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108648000B (en) | 2022-10-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108648000A (en) | Method and device, the electronic equipment that life cycle is assessed are retained to user | |
US20240165484A1 (en) | Method and system for interactive, interpretable, and improved match and player performance predictions in team sports | |
Miljković et al. | The use of data mining for basketball matches outcomes prediction | |
US11726900B2 (en) | Optimized test case selection for quality assurance testing of video games | |
CN108122127A (en) | Predict the method and device of the operation result of game on line service | |
CN107335220B (en) | Negative user identification method and device and server | |
Danisik et al. | Football match prediction using players attributes | |
CN108066989A (en) | A kind of random fit organizing method, device and application server | |
Zare et al. | Engineering features to improve pass prediction in soccer simulation 2d games | |
US20240181343A1 (en) | System and method for individual player and team simulation | |
Freund | Computer and information sciences | |
Wang et al. | Match tracing: A unified framework for real-time win prediction and quantifiable performance evaluation | |
CN109499069A (en) | Operating result method of calibration, device, storage medium and electronic device | |
Eger et al. | Deck archetype prediction in hearthstone | |
Kröckel et al. | Process mining of football event data: a novel approach for tactical insights into the game | |
Robberechts et al. | un-xPass: Measuring Soccer Player's Creativity | |
Ramalingam et al. | Efficient Player Prediction and Suggestion using Machine Learning for IPL Tournament | |
CN110941769A (en) | Target account determination method and device and electronic device | |
CN112818241B (en) | Content promotion method and device, computer equipment and storage medium | |
Hamdad et al. | Basketball analytics. Data mining for acquiring performances | |
CN113971582A (en) | Method and system for generating advertisement putting plan, electronic device and storage medium | |
Rahaie et al. | Critic learning in multi agent credit assignment problem | |
Elbert | Process and Strategy Mining in Real-Time Strategy Games | |
CN110443374A (en) | A kind of resource information processing method, device and equipment | |
Kwon et al. | Clustering player behavioral data and improving performance of churn prediction from mobile game |
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 |