CN107682178A - A kind of mobile subscriber's online operation behavior Forecasting Methodology and device - Google Patents
A kind of mobile subscriber's online operation behavior Forecasting Methodology and device Download PDFInfo
- Publication number
- CN107682178A CN107682178A CN201710764731.7A CN201710764731A CN107682178A CN 107682178 A CN107682178 A CN 107682178A CN 201710764731 A CN201710764731 A CN 201710764731A CN 107682178 A CN107682178 A CN 107682178A
- Authority
- CN
- China
- Prior art keywords
- data
- decision
- mobile subscriber
- operation behavior
- online
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/147—Network analysis or design for predicting network behaviour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
- G06F16/972—Access to data in other repository systems, e.g. legacy data or dynamic Web page generation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of mobile subscriber's online operation behavior Forecasting Methodology and device, including：Obtain the data set of original mobile subscriber's online operation behavior；Pretreatment acquisition training set is carried out to the data of data set；According to decision tree principle and default division rule, determine attribute corresponding to decision tree different levels and lowest hierarchical level corresponding to classification, establish decision-tree model；Attribute corresponding to the first level is the online position of mobile subscriber in decision-tree model；According to property value corresponding with each attribute in decision-tree model in training set data and mobile subscriber corresponding with classification online operation behavior data, training set data input decision-tree model is trained；The data as mode input in the behavioral data of certain specific mobile subscriber are obtained, by decision-tree model, next step behavior is predicted.The online operation behavior of the simple data prediction cellphone subscriber such as positional information can be utilized, and then adaptable personalized service is provided for different upper net operation clients.
Description
Technical field
The present invention relates to user's online operation behavior Forecasting Methodology, and in particular to one kind utilizes the prediction movement of decision tree principle
The method and apparatus of user's online operation behavior.
Background technology
The development of mobile Internet was swift and violent in recent years, and the 3G/4G networks of high speed are increasingly popularized, according to China Internet network
The report of the newest phase of information centre (CNNIC), by June, 2017, China's netizen's scale reaches 7.51 hundred million, and wherein mobile phone moves
For the scale of dynamic netizen up to 7.24 hundred million, the ratio for accounting for overall netizen reaches 96.3%.And as mobile Internet further develops, thing
The further popularization of the surfing Internets with cell phone such as networking application, the popular degree of dependence to surfing Internet with cell phone can more and more highers.
In face of so huge mobile phone user, how by the data such as big data and user's custom and using suitable
Method prejudge pattern and the concrete operations of mobile subscriber's online, so as to providing properer and personalized movement for client
The problem that Internet service is the communications field all the time and internet arena is paid special attention to.
For example, in the past according to the type of cell-phone number and operator's (such as Global Link), network access (such as 3G, 2G), mobile phone product
Board (such as apple, millet) data substantially predict that network access, such as prediction Global Link client can carry out high-end consumption and provide
Related information, or prediction millet cell phone customer are that young man is in the majority, and the possibility that game is played in online is larger, so as to provide more
The related information content of more game.
In addition, in a kind of patent publication " user gender prediction method based on surfing Internet with cell phone operation behavior " (invention
Number of patent application 201611127122.2) in refer to the sex clicking on related data according to APP and judge surfing Internet with cell phone user,
So as to provide the technical scheme supported to subsequently carrying out related personalized service recommendation according to the preference of different sexes user.
But collection of the conventional method to the data of cellphone subscriber excessively sectionalization, universality is not embodied, is not also had
Have and provide specifically how influence to online operation behavior is sentenced to the significant online place of online operation behavior
Disconnected method, therefore under the increasingly huger background of the dimension of mobile Internet user data, do not provide more practicability and effectiveness
Mobile subscriber surf the Net operation behavior Forecasting Methodology.
The content of the invention
To solve above-mentioned problems of the prior art, the present inventor, which has paid close attention to, has arrived online place
Influence to operation behavior of surfing the Net is very big, and decision-making is carried out between online place and online operation behavior using decision-making tree theory
Match somebody with somebody, so as to predict the online operation behavior of cellphone subscriber according to simple master data.
Herein, surf the Net place include in general at home, Internet bar, unit, school and public place etc.；And surf the Net
Operation behavior is according to above-mentioned CNNIC statistics, from more to less including mobile phone instant messaging, cell phone network news, mobile phone searching, hand
Machine online music, mobile phone online payment, cell phone network shopping, mobile phone network game, mobile phone Web bank, cellular network roadbed text
, mobile telephone for tourist reservation, mobile E-mail, mobile phone forum, mobile phone online education course, mobile phone microblogging, cell phone map navigation, mobile phone
Suscribe to take-away etc. on the net.
The present invention specifically provides a kind of mobile subscriber's online operation behavior Forecasting Methodology and device, it is characterised in that including
Following steps：Obtain the data set of original mobile subscriber's online operation behavior；The data of the data set are pre-processed
Obtain training set；According to decision tree principle and default division rule, attribute corresponding to decision tree different levels and most is determined
Classification corresponding to low-level, establishes decision-tree model, wherein, the attribute is used to divide data in corresponding level；
In the decision-tree model, attribute corresponding to the first level is the online position of mobile subscriber；According in training set data with institute
Property value corresponding to each attribute in decision-tree model, and mobile subscriber corresponding with classification online operation behavior data are stated, will
The training set data inputs the decision-tree model and the decision-tree model is trained；Obtain certain specific mobile subscriber's
Data in behavioral data as mode input, by above-mentioned decision-tree model, the next step behavior to mobile subscriber carries out pre-
Survey.
Preferred embodiment is that attribute corresponding to the level of decision tree second includes indoor and outdoors；Outside the decision tree room
Third layer level attribute corresponding to branch is included in walking about and in public transit facility is taken.
Preferred embodiment is that the division rule of foundation includes when being divided based on the second level attributes to data：Work as regulation
Average moving distance in time is judged as when being less than or equal to pre-determined distance threshold value indoors, when more than pre-determined distance threshold value
It is judged as in outdoor；The division rule of foundation includes when being divided based on third layer level attribute to data：When being judged as outdoor
And the average translational speed in the stipulated time is less than or equal to be judged as walking during the first pre-set velocity, when being judged as outdoor and advise
Average translational speed in fixing time is judged as in public transit facility is taken when being more than the second pre-set velocity；Wherein, it is described
After average moving distance is obtains current location with specific time interval at the appointed time, current location and before most is determined
The distance between nearly position once determined, and multiple distances within the stipulated time are subjected to the distance after being averaged.
Further mode of priority is that the stipulated time is 30 minutes, and the predetermined distance is 5 minutes.
Mode of priority is that mobile subscriber's online operation behavior includes web page browsing, shopping online and online game.
Further mode of priority is that the training set data is inputted into the decision-tree model enters to the decision-tree model
Row training, is specifically included：The training set data is inputted into the decision-tree model, for each online position is divided in decision tree
Mobile subscriber's online operation behavior corresponding to lowest hierarchical level in branch, according on this in the training set of mobile subscriber's online operation behavior
The data of net position correspondence, determine each online operation behavior probability of occurrence；Probability of occurrence according to corresponding online operation behavior
Calculate the weight coefficient of corresponding online operation behavior；Obtain the number as mode input in the behavioral data of certain specific mobile subscriber
According to by above-mentioned decision-tree model, the next step behavior to mobile subscriber is predicted, specifically included：Obtain certain specific movement
Data in the behavioral data of user as mode input, and input the decision-tree model；When the lowest hierarchical level in decision tree
Last layer level when being judged, the maximum upper net operation row of weight coefficient in the online operation behavior that lowest hierarchical level is included
To be defined as the next step behavior of the mobile subscriber of prediction.
Preferred embodiment is, after the next step behavior to mobile subscriber is predicted, in addition to：Collect the mobile use
The feedback information of the real next step behavior in family；Prediction result and feedback information are contrasted；For prediction result and feedback
The incongruent situation of information, the branch that father node where the mobile subscriber is corresponded into prediction result described in decision tree is included
In, the weight coefficient of operation behavior of surfing the Net corresponding with the feedback information is heightened.
The present invention also provides a kind of mobile subscriber's online operation behavior prediction meanss, it is characterised in that including：Initial data
Collect acquiring unit, obtain the data set of original mobile subscriber's online operation behavior；Training set acquiring unit, to the data set
Data carry out pretreatment obtain training set；Decision-tree model establishes unit, according to decision tree principle and default division rule,
Classification corresponding to determining attribute corresponding to decision tree different levels and lowest hierarchical level, establishes decision-tree model, wherein, the category
Property be used for data are divided in corresponding level；In the decision-tree model, attribute corresponding to the first level is used to be mobile
The online position at family；Decision-tree model training unit, according to corresponding with each attribute in the decision-tree model in training set data
Property value, and mobile subscriber corresponding with classification surfs the Net operation behavior data, and training set data input is described certainly
Plan tree-model is trained to the decision-tree model；Operation predicting unit in next step, obtains the behavior of certain specific mobile subscriber
Data in data as mode input, by above-mentioned decision-tree model, the next step behavior to mobile subscriber is predicted.It is logical
Technical scheme is crossed, the simple data such as positional information can be utilized, using decision-making tree theory, predict cellphone subscriber's
Online operation behavior, so as to provide adaptable personalized service for different upper net operation clients.
Brief description of the drawings
Fig. 1 represents the schematic flow sheet of mobile subscriber's online operation behavior Forecasting Methodology of the present invention.
Fig. 2 represents the prediction schematic diagram of the operation behavior using one embodiment of the present of invention mobile Internet access user.
Embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is the present invention
Part of the embodiment, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not having
The every other embodiment obtained under the premise of creative work is made, belongs to the scope of protection of the invention.
Below, referring to the drawings, the embodiment of the present invention is described in detail.
Fig. 1 represents the schematic flow sheet of mobile subscriber's online operation behavior Forecasting Methodology of the present invention.As shown in figure 1, this
Mobile subscriber's online operation behavior Forecasting Methodology of invention comprises the following steps：
(1) data set of original mobile subscriber's online operation behavior is obtained；
(2) pretreatment acquisition training set is carried out to the data of the data set；
(3) according to decision tree principle and default division rule, attribute corresponding to decision tree different levels and most is determined
Classification corresponding to low-level, establishes decision-tree model, wherein, the attribute is used to divide data in corresponding level；
In the decision-tree model, attribute corresponding to the first level is the online position of mobile subscriber；
(4) according to property value corresponding with each attribute in the decision-tree model in training set data, and with classification pair
The mobile subscriber's online operation behavior data answered, the decision-tree model is inputted to the decision tree mould by the training set data
Type is trained；
(5) data as mode input in the behavioral data of certain specific mobile subscriber are obtained, pass through above-mentioned decision tree mould
Type, the next step behavior to mobile subscriber are predicted.
Each step is described in detail separately below.
(1) data set of original mobile subscriber's online operation behavior is obtained
The data set of original mobile subscriber's online operation behavior is collected first, according to each attribute, obtains each mobile use
The attributes such as the average moving distance at the appointed time at family, the average translational speed in the stipulated time and specifically upper net operation
The corresponding data of behavior.
Then the form by above-mentioned data creating into a form, wherein, often row represents different users, and each column represents not
Same attribute, last row is particularly net operation behavior, that is, corresponds to classification.The file of the form becomes initial
Data set.
Wherein, in above-mentioned attribute can also including the use of service type (such as Global Link), sex, data traffic size
Deng other data.
Wherein, final mobile subscriber's operation behavior can include mobile phone instant messaging, cell phone network news, mobile phone searching,
Cell phone network music, mobile phone online payment, cell phone network shopping, mobile phone network game, mobile phone Web bank, cellular network roadbed text
, mobile telephone for tourist reservation, mobile E-mail, mobile phone forum, mobile phone online education course, mobile phone microblogging, cell phone map navigation, mobile phone
Suscribe to take-away etc. on the net.In these mobile subscriber's operation behaviors, it can be sorted out or screened with only a part operation interested,
For example, web page browsing, shopping online and online game etc..
Wherein, after the average moving distance is obtains current location with specific time interval at the appointed time, survey
Settled front position and the distance between position of the last measure, and multiple distances within the stipulated time are entered before
Distance after row is average.Stipulated time and the stipulated time time interval can freely be set according to actual conditions, for example,
The stipulated time can be 10 minutes, 20 minutes, 30 minutes or one hours, and the predetermined distance can be 1 minute, 2
Minute, 5 minutes or 10 minutes.
(2) pretreatment acquisition training set is carried out to the data of the data set
Next the data set for operation behavior of being surfed the Net to the original mobile subscriber pre-processes.That is, deletion need not
The attribute column wanted, and unconcerned surfing Internet with cell phone behavior.For example the attribute column of Information Mobile Service type can be deleted, and delete
Except the row for listening attentively to cell phone network music.And the infull sample of data can also be deleted, so as to which acquisition carries out pretreatment simplification
Data set afterwards.
Then, a part will be extracted in the data set after handling as training set, the foundation for decision model.
(3) according to decision tree principle and default division rule, attribute corresponding to decision tree different levels and most is determined
Classification corresponding to low-level, establishes decision-tree model, wherein, the attribute is used to divide data in corresponding level；
In the decision-tree model, attribute corresponding to the first level is the online position of mobile subscriber.
The online position includes indoor and outdoors, wherein outdoor also include in walking about and taking public transit facility
In.That is, stated according to decision tree principle, attribute corresponding to the level of decision tree second includes indoor and outdoors, should be based on the
The division rule of foundation includes when two level attributes divide to data：Be less than when the average moving distance in the stipulated time or
It is judged as indoors, when more than pre-determined distance threshold value being judged as in outdoor during equal to pre-determined distance threshold value.
For example, it is judged as indoors when the average moving distance in the stipulated time is less than or equal to 30 meters, when big
It is judged as when 30 meters in outdoor.Wherein, the average moving distance is worked as to be obtained at the appointed time with specific time interval
After front position, current location and the distance between position of the last measure, and will be in the stipulated time before are determined
Interior multiple distances carry out the distance after being averaged.
In addition, third layer level attribute corresponding to the decision tree outdoor branch includes in walking about and is taking public transport
In facility.
That is, stated according to decision tree principle, the division rule of foundation when being divided based on third layer level attribute to data
Including：It is judged as being expert at when the average translational speed being judged as in the outdoor and stipulated time is less than or equal to the first pre-set velocity
Walk, be judged as taking public friendship when the average translational speed being judged as in the outdoor and stipulated time is more than the second pre-set velocity
In logical facility.
For example, it is judged as when the average translational speed being judged as in the outdoor and stipulated time is less than or equal to 2 meter per second
Walking, be judged as taking public transport when the average translational speed being judged as in the outdoor and stipulated time is more than 5 meter per second
In facility.
In addition, the stipulated time preferably can be 30 minutes, the predetermined distance preferably can be 5 minutes.The shifting
Web page browsing, shopping online and online game can preferably be included by employing family online operation behavior.Herein, to training set
Data are further processed, i.e. the data for the training set of operation behavior of being surfed the Net according to mobile subscriber, to each point of decision tree
Each operation behavior under branch calculates weight coefficient according to probability of occurrence.
(4) according to property value corresponding with each attribute in the decision-tree model in training set data, and with classification pair
The mobile subscriber's online operation behavior data answered, the decision-tree model is inputted to the decision tree mould by the training set data
Type is trained.
The training specifically includes：
(a) training set data is inputted into the decision-tree model, in position branch of each being surfed the Net in decision tree
Mobile subscriber's online operation behavior corresponding to lowest hierarchical level, according to the online position in the training set of mobile subscriber's online operation behavior
Corresponding data are put, determine each online operation behavior probability of occurrence；
(b) probability of occurrence according to corresponding online operation behavior calculates the weight coefficient of corresponding online operation behavior；
(c) data as mode input in the behavioral data of certain specific mobile subscriber are obtained, pass through above-mentioned decision tree mould
Type, the next step behavior to mobile subscriber are predicted, specifically included：
(d) data as mode input in the behavioral data of certain specific mobile subscriber are obtained, and input the decision tree
Model；
(e) when the last layer level of the lowest hierarchical level in decision tree is judged, upper net operation that lowest hierarchical level is included
The maximum online operation behavior of weight coefficient in behavior, is defined as the next step behavior of the mobile subscriber of prediction.
In addition, after the next step behavior to mobile subscriber is predicted, in addition to：It is true to collect the mobile subscriber
Next step behavior feedback information；Prediction result and feedback information are contrasted；For prediction result and feedback information not
Situation about meeting, the mobile subscriber is corresponded in the branch that the place of prediction result described in decision tree father node is included, with
Correspondingly the surf the Net weight coefficient of operation behavior of the feedback information is heightened.
(5) data as mode input in the behavioral data of certain specific mobile subscriber are obtained, pass through above-mentioned decision tree mould
Type, the next step behavior to mobile subscriber is predicted after decision tree after above-mentioned training is obtained, to certain specific user
Operation behavior when being predicted, the data as mode input are extracted in the behavioral data of the specific mobile subscriber, then
Judge through above-mentioned decision-tree model, the next step operation behavior of the final mobile subscriber for obtaining prediction.
After the next step operation behavior of the mobile subscriber is predicted, it can be provided according to the predictive behavior as suitable
The adaptable personalized service such as information.
Mobile subscriber's online operation behavior Forecasting Methodology of the invention described above can be realized by following device, i.e. the present invention
Mobile subscriber surf the Net operation behavior prediction meanss, including：Raw data set acquiring unit, obtain original mobile subscriber's online
The data set of operation behavior；Training set acquiring unit, pretreatment acquisition training set is carried out to the data of the data set；Decision tree
Model establishes unit, according to decision tree principle and default division rule, determine attribute corresponding to decision tree different levels and
Classification corresponding to lowest hierarchical level, establishes decision-tree model, wherein, the attribute is used to divide data in corresponding level；
In the decision-tree model, attribute corresponding to the first level is the online position of mobile subscriber；Decision-tree model training unit,
According to property value corresponding with each attribute in the decision-tree model in training set data, and mobile subscriber corresponding with classification
Online operation behavior data, the training set data is inputted into the decision-tree model decision-tree model is trained；
Operation predicting unit in next step, obtains the data as mode input in the behavioral data of certain specific mobile subscriber, by above-mentioned
Decision-tree model, the next step behavior to mobile subscriber are predicted.
Further, attribute corresponding to the level of decision tree second includes indoor and outdoors；Divide outside the decision tree room
Third layer level attribute corresponding to branch is included in walking about and in public transit facility is taken.
Further, the division rule of foundation includes when being divided based on the second level attributes to data：When regulation
Interior average moving distance is judged as indoors, when more than pre-determined distance threshold value sentencing when being less than or equal to pre-determined distance threshold value
Break as in outdoor；The division rule of foundation includes when being divided based on third layer level attribute to data：When be judged as it is outdoor and
Average translational speed in stipulated time is less than or equal to be judged as walking during the first pre-set velocity, when being judged as outdoor and provide
Average translational speed in time is judged as in public transit facility is taken when being more than the second pre-set velocity；Wherein, it is described flat
After equal displacement is obtains current location with specific time interval at the appointed time, current location and nearest before is determined
The distance between position once determined, and multiple distances within the stipulated time are subjected to the distance after being averaged.
Further, the stipulated time is 30 minutes, and the predetermined distance is 5 minutes.
Further, the decision-tree model training unit, specifically for the training set data is inputted into the decision-making
Tree-model, in decision tree each online position branch in corresponding to lowest hierarchical level mobile subscriber surf the Net operation behavior, according to
The data of the online position correspondence in the training set of mobile subscriber's online operation behavior, it is general to determine that each online operation behavior occurs
Rate；Probability of occurrence according to corresponding online operation behavior calculates the weight coefficient of corresponding online operation behavior；Obtain certain specific shifting
The data as mode input in the behavioral data at family are employed, by above-mentioned decision-tree model, next walking to mobile subscriber
To be predicted, specifically include：The data as mode input in the behavioral data of certain specific mobile subscriber are obtained, and input institute
State decision-tree model；When the last layer level of the lowest hierarchical level in decision tree is judged, the online that lowest hierarchical level is included is grasped
Make the online operation behavior that weight coefficient is maximum in behavior, be defined as the next step behavior of the mobile subscriber of prediction.
Further, it is described also to include：Feedback unit；The feedback unit, for operating predicting unit pair in next step
After the next step behavior of mobile subscriber is predicted, the feedback information of the real next step behavior of the mobile subscriber is collected；
Prediction result and feedback information are contrasted；For prediction result and the incongruent situation of feedback information, by the mobile use
In the branch that father node where family corresponds to prediction result described in decision tree is included, upper net operation corresponding with the feedback information
The weight coefficient of behavior is heightened.
(embodiment)
Fig. 2 represents the prediction schematic diagram of the operation behavior using one embodiment of the present of invention mobile Internet access user.Below,
With reference to figure 2, embodiments of the invention are described in detail.
The data set of original mobile subscriber's online operation behavior is obtained, is surfed the Net according to original mobile subscriber is collected first
The data set of operation behavior, according to each attribute, average moving distance at the appointed time, the average movement in the stipulated time
Speed, particularly net operation behavior carry out data taxonomic revision.
Form is made in the data of above-mentioned attribute and final operation behavior, as initial data set.Wherein, upper net operation
Behavior only selects three kinds of web page browsing, shopping online and online game.
In addition, after the average moving distance is obtains current location with specific time interval at the appointed time, survey
The distance between settled front position and the position that determines before, and by multiple distances within the stipulated time carry out it is average it
Distance afterwards.The stipulated time is set herein as 30 minutes, and the predetermined distance is 5 minutes.
Next, the data set for operation behavior of being surfed the Net to the original mobile subscriber pre-processes.That is, webpage is deleted
Browse, the data that shopping online and online game three behaviors are unrelated, and whole data set is simplified.
Then, a part will be extracted in the data set after handling as training set, the foundation for decision model.
The online position includes indoor and outdoors, wherein outdoor also include in walking about and taking public transit facility
In.
Next, establish decision-tree model with training set data.In the present embodiment, the first order is divided according to position
Class, i.e. be judged as indoors, when more than 30 meters judging when the average moving distance in the stipulated time is less than or equal to 30 meters
For in outdoor.
In the classification of the second level, situation indoors is categorized as three kinds of operation behaviors, i.e. web page browsing, open game, net
Upper shopping.A weight coefficient is assigned to each online operation behavior, web page browsing A1, opening game are B1, online purchase
Thing is C1, and these weight coefficients are to carry out probabilistic operation acquisition to data in the data set after above-mentioned simplification.For example, in number
If the data in the case of according to concentrating indoors assume there are 100, and the behavior of wherein web page browsing has 50, opens game
Behavior have 30, the behavior of shopping online has 20, then weight coefficient is 0.5,0.3 and 0.2 respectively.
In the case of in outdoor, separated further according to translational speed.That is, when being judged as in the outdoor and stipulated time
Average translational speed is less than or equal to be judged as walking during 2 meter per second, when the average mobile speed being judged as in the outdoor and stipulated time
Degree is judged as in public transit facility is taken when being more than 5 meter per second.
Speed is more than 2 meter per seconds and when being less than or equal to 5 meter per second, and the speed is to be unlikely to be speed on foot, also not
Be probably very much take bus, the public transport such as train, it is likely in cycling etc., therefore carry out web page browsing, open
The possibility very little of the behaviors such as game, shopping online, so being directly judged to not surfing the Net.
In the case of translational speed is less than or equal to 2 meter per seconds and is judged as in walking, further according to the number after above-mentioned simplification
The weight coefficient that data are carried out with probabilistic operation acquisition and is directed to web page browsing, opening game, shopping online according to concentrating, i.e. webpage
Browse for A2, open game be B2, shopping online C2.
It is judged as taking the situation of public transit facility similarly, for 5 meter per seconds are more than in translational speed,
Imparting web page browsing is A3, opening game is B3, the weight coefficient that shopping online is C3.
Herein, A1, B1, C1 and A2, B2, C2 and A3, B3, C3 are according to the different same upper net operation row of residing environment
Correlation for corresponding weight coefficient is also not quite similar.Such as when walking about compared with the probability phase for indoors, opening game
To low, i.e. B1 ＞ B2.
Which kind of when carrying out decision tree judgement, when needing to judge carrying out operation under certain state, compare
A, the size between B, C, using one of value maximum as the operation behavior finally judged.One volume can also be set herein
Outer condition, i.e. only the value is maximum and is sentenced more than ability more than certain numerical value, such as 0.5 as final operation behavior
It is disconnected.
Next, for obtaining the behavioral data of certain specific mobile subscriber as sample, with the normal of the mobile subscriber
The input as model is inputted, by above-mentioned decision-tree model, the next step behavior to mobile subscriber is predicted.
When being predicted, corresponding weight coefficient (A, B, C) can be entered according to the specific correction behavioral data of the user
Row adjustment, and user's internet behavior is predicted based on the weight coefficient after adjustment, wherein, the specific correction behavior number of user
The operation behavior interested according to user is characterized.
For example, know that certain specific user likes opening game (correction behavioral data) by historical data, then according to correction
Data are adjusted to B1, such as 0.1, then B ' 1=B1 (0.3)+0.1=0.4, then entered with the data after the correction with A1 and C1
Row compares.The value of the adjustment can be a fixed numbers or carry out various changes according to residing different state
Change.
Further, because B1 is changed, in order to ensure A1, B1 and C1 and for 1, the value of three can be entered
Row processing, is repeated no more here.
Above to the present invention embodiment and embodiment be described in detail, but those skilled in the art according to
The scope that foregoing description can be changed and replaced falls within the protection content of the present invention.
For example, three kinds of operation behaviors of the present embodiment can expand to more.
For example, the lower section of the operation behavior of the present embodiment can also increase multiclass classification, for example, indoors when be judged as net
During page browsing, continue web page browsing after certain time or the possibility of opening game or shopping online also can
Change, can also now be determined whether to this part.That is, decision tree judges with infinite stages to extend in theory.
Described above is only the preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, it can also make and be suitably modified and deform, these are improved and deformation
Also it should be regarded as protection scope of the present invention.
Claims (10)
- The operation behavior Forecasting Methodology 1. a kind of mobile subscriber surfs the Net, it is characterised in that comprise the following steps：Obtain the data set of original mobile subscriber's online operation behavior；Pretreatment acquisition training set is carried out to the data of the data set；According to decision tree principle and default division rule, attribute and lowest hierarchical level pair corresponding to decision tree different levels are determined The classification answered, establishes decision-tree model, wherein, the attribute is used to divide data in corresponding level；In the decision-making In tree-model, attribute corresponding to the first level is the online position of mobile subscriber；According to property value corresponding with each attribute in the decision-tree model in training set data, and corresponding with classification move User's online operation behavior data, the training set data is inputted into the decision-tree model decision-tree model is instructed Practice；The data as mode input in the behavioral data of certain specific mobile subscriber are obtained, by above-mentioned decision-tree model, to moving The next step behavior for employing family is predicted.
- The operation behavior Forecasting Methodology 2. mobile subscriber according to claim 1 surfs the Net, it is characterised in that the decision tree the Attribute corresponding to two levels includes indoor and outdoors；Third layer level attribute corresponding to the decision tree outdoor branch is included in walking about And in public transit facility is taken.
- The operation behavior Forecasting Methodology 3. mobile subscriber according to claim 1 surfs the Net, it is characterised in that based on the second level The division rule of foundation includes when attribute divides to data：When the average moving distance in the stipulated time is less than or equal in advance If it is judged as indoors, when more than pre-determined distance threshold value being judged as in outdoor during distance threshold；The division rule of foundation includes when being divided based on third layer level attribute to data：When being judged as the outdoor and stipulated time Interior average translational speed is less than or equal to be judged as walking during the first pre-set velocity, when being judged as in the outdoor and stipulated time Average translational speed is judged as in public transit facility is taken when being more than the second pre-set velocity；Wherein, after the average moving distance is obtains current location with specific time interval at the appointed time, measure is worked as Front position and the distance between position of the last measure, and multiple distances within the stipulated time are put down before Distance after.
- The operation behavior Forecasting Methodology 4. mobile subscriber according to claim 3 surfs the Net, it is characterised in thatThe stipulated time is 30 minutes, and the predetermined distance is 5 minutes.
- The operation behavior Forecasting Methodology 5. mobile subscriber according to claim 1 surfs the Net, it is characterised in thatMobile subscriber's online operation behavior includes web page browsing, shopping online and online game.
- The operation behavior Forecasting Methodology 6. mobile subscriber according to claim 5 surfs the Net, it is characterised in that by the training set Decision-tree model described in data input is trained to the decision-tree model, is specifically included：The training set data is inputted into the decision-tree model, for lowest hierarchical level in position branch of each being surfed the Net in decision tree Corresponding mobile subscriber's online operation behavior, according to the online position correspondence in the training set of mobile subscriber's online operation behavior Data, determine each online operation behavior probability of occurrence；Probability of occurrence according to corresponding online operation behavior calculates the weight coefficient of corresponding online operation behavior；The data as mode input in the behavioral data of certain specific mobile subscriber are obtained, by above-mentioned decision-tree model, to moving The next step behavior for employing family is predicted, and is specifically included：The data as mode input in the behavioral data of certain specific mobile subscriber are obtained, and input the decision-tree model；When the last layer level of the lowest hierarchical level in decision tree is judged, add in the online operation behavior that lowest hierarchical level is included The maximum online operation behavior of weight coefficient, is defined as the next step behavior of the mobile subscriber of prediction.
- The operation behavior Forecasting Methodology 7. mobile subscriber according to claim 1 surfs the Net, it is characterised in that to mobile subscriber Next step behavior be predicted after, in addition to：Collect the feedback information of the real next step behavior of the mobile subscriber；Prediction result and feedback information are contrasted；For prediction result and the incongruent situation of feedback information, the mobile subscriber is corresponded into prediction result described in decision tree In the branch that place father node is included, the weight coefficient of operation behavior of surfing the Net corresponding with the feedback information is heightened.
- The operation behavior prediction meanss 8. a kind of mobile subscriber surfs the Net, it is characterised in that including：Raw data set acquiring unit, obtain the data set of original mobile subscriber's online operation behavior；Training set acquiring unit, pretreatment acquisition training set is carried out to the data of the data set；Decision-tree model establishes unit, according to decision tree principle and default division rule, determines that decision tree different levels are corresponding Attribute and lowest hierarchical level corresponding to classification, establish decision-tree model, wherein, the attribute be used in corresponding level to data Divided；In the decision-tree model, attribute corresponding to the first level is the online position of mobile subscriber；Decision-tree model training unit, according to property value corresponding with each attribute in the decision-tree model in training set data, And mobile subscriber's online operation behavior data corresponding with classification, the training set data is inputted into the decision-tree model pair The decision-tree model is trained；Operation predicting unit in next step, obtains the data as mode input in the behavioral data of certain specific mobile subscriber, passes through Above-mentioned decision-tree model, the next step behavior to mobile subscriber are predicted.
- The operation behavior prediction meanss 9. mobile subscriber according to claim 8 surfs the Net, it is characterised in that the decision tree the Attribute corresponding to two levels includes indoor and outdoors；Third layer level attribute corresponding to the decision tree outdoor branch is included in walking about And in public transit facility is taken.
- The operation behavior prediction meanss 10. mobile subscriber according to claim 8 surfs the Net, it is characterised in that based on the second layer The division rule of foundation includes when level attribute divides to data：When the average moving distance in the stipulated time is less than or equal to It is judged as indoors, when more than pre-determined distance threshold value being judged as in outdoor during pre-determined distance threshold value；The division rule of foundation includes when being divided based on third layer level attribute to data：When being judged as the outdoor and stipulated time Interior average translational speed is less than or equal to be judged as walking during the first pre-set velocity, when being judged as in the outdoor and stipulated time Average translational speed is judged as in public transit facility is taken when being more than the second pre-set velocity；Wherein, after the average moving distance is obtains current location with specific time interval at the appointed time, measure is worked as Front position and the distance between position of the last measure, and multiple distances within the stipulated time are put down before Distance after.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710764731.7A CN107682178A (en) | 2017-08-30 | 2017-08-30 | A kind of mobile subscriber's online operation behavior Forecasting Methodology and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710764731.7A CN107682178A (en) | 2017-08-30 | 2017-08-30 | A kind of mobile subscriber's online operation behavior Forecasting Methodology and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107682178A true CN107682178A (en) | 2018-02-09 |
Family
ID=61134488
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710764731.7A Pending CN107682178A (en) | 2017-08-30 | 2017-08-30 | A kind of mobile subscriber's online operation behavior Forecasting Methodology and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107682178A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109800359A (en) * | 2018-12-20 | 2019-05-24 | 北京百度网讯科技有限公司 | Information recommendation processing method, device, electronic equipment and readable storage medium storing program for executing |
| CN109978627A (en) * | 2019-03-29 | 2019-07-05 | 电子科技大学中山学院 | Modeling method for big data of user internet access behavior of broadband access network |
| CN110303531A (en) * | 2018-03-27 | 2019-10-08 | 博朗有限公司 | Personal care device and its operating method |
| CN111523032A (en) * | 2020-04-22 | 2020-08-11 | 贝壳技术有限公司 | Method, device, medium and electronic equipment for determining user preference |
| CN112580902A (en) * | 2021-02-26 | 2021-03-30 | 腾讯科技（深圳）有限公司 | Object data processing method and device, computer equipment and storage medium |
| US11298839B2 (en) | 2018-03-27 | 2022-04-12 | Braun Gmbh | Hair removal device |
| US11548177B2 (en) | 2018-03-27 | 2023-01-10 | Braun Gmbh | Personal care device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102238045A (en) * | 2010-04-27 | 2011-11-09 | 广州迈联计算机科技有限公司 | System and method for predicting user behavior in wireless Internet |
| KR101312446B1 (en) * | 2011-11-25 | 2013-09-27 | 주식회사 포커스원 | Analysis apparatus and method for usability of mobile application using user bebavior log |
| CN104111920A (en) * | 2013-04-16 | 2014-10-22 | 华为技术有限公司 | Decision-making tree based prediction method and device |
| CN104331404A (en) * | 2013-07-22 | 2015-02-04 | 中国科学院深圳先进技术研究院 | A user behavior predicting method and device based on net surfing data of a user's cell phone |
| CN104468257A (en) * | 2014-10-30 | 2015-03-25 | 北京大学 | Cloud application availability prediction method and system based on mobile user time-space behaviors |
| CN106408343A (en) * | 2016-09-23 | 2017-02-15 | 广州李子网络科技有限公司 | Modeling method and device for user behavior analysis and prediction based on BP neural network |
| CN106991496A (en) * | 2017-03-29 | 2017-07-28 | 南京邮电大学 | A kind of user behavior towards mobile social environment is layered interaction prediction method |
-
2017
- 2017-08-30 CN CN201710764731.7A patent/CN107682178A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102238045A (en) * | 2010-04-27 | 2011-11-09 | 广州迈联计算机科技有限公司 | System and method for predicting user behavior in wireless Internet |
| KR101312446B1 (en) * | 2011-11-25 | 2013-09-27 | 주식회사 포커스원 | Analysis apparatus and method for usability of mobile application using user bebavior log |
| CN104111920A (en) * | 2013-04-16 | 2014-10-22 | 华为技术有限公司 | Decision-making tree based prediction method and device |
| CN104331404A (en) * | 2013-07-22 | 2015-02-04 | 中国科学院深圳先进技术研究院 | A user behavior predicting method and device based on net surfing data of a user's cell phone |
| CN104468257A (en) * | 2014-10-30 | 2015-03-25 | 北京大学 | Cloud application availability prediction method and system based on mobile user time-space behaviors |
| CN106408343A (en) * | 2016-09-23 | 2017-02-15 | 广州李子网络科技有限公司 | Modeling method and device for user behavior analysis and prediction based on BP neural network |
| CN106991496A (en) * | 2017-03-29 | 2017-07-28 | 南京邮电大学 | A kind of user behavior towards mobile social environment is layered interaction prediction method |
Non-Patent Citations (1)
| Title |
|---|
| 罗海艳: ""移动用户网络行为分析与预测方法研究"", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110303531A (en) * | 2018-03-27 | 2019-10-08 | 博朗有限公司 | Personal care device and its operating method |
| US11298839B2 (en) | 2018-03-27 | 2022-04-12 | Braun Gmbh | Hair removal device |
| US11548177B2 (en) | 2018-03-27 | 2023-01-10 | Braun Gmbh | Personal care device |
| CN109800359A (en) * | 2018-12-20 | 2019-05-24 | 北京百度网讯科技有限公司 | Information recommendation processing method, device, electronic equipment and readable storage medium storing program for executing |
| CN109978627A (en) * | 2019-03-29 | 2019-07-05 | 电子科技大学中山学院 | Modeling method for big data of user internet access behavior of broadband access network |
| CN109978627B (en) * | 2019-03-29 | 2023-08-08 | 电子科技大学中山学院 | Modeling method for big data of broadband access network user surfing behavior |
| CN111523032A (en) * | 2020-04-22 | 2020-08-11 | 贝壳技术有限公司 | Method, device, medium and electronic equipment for determining user preference |
| CN112580902A (en) * | 2021-02-26 | 2021-03-30 | 腾讯科技（深圳）有限公司 | Object data processing method and device, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107682178A (en) | A kind of mobile subscriber's online operation behavior Forecasting Methodology and device | |
| Lee et al. | MONERS: A news recommender for the mobile web | |
| KR101418393B1 (en) | Apparatus and method for mobile intelligent advertizing based on mobile user contextual matching | |
| CN102033883A (en) | Method, device and system for improving data transmission speed of website | |
| CN104156897B (en) | Indoor guide system based on context aware | |
| CN103246725A (en) | Wireless network based data traffic pushing system and method | |
| CN103235823A (en) | Method and system for determining current interest of users according to related web pages and current behaviors | |
| CN103235824A (en) | Method and system for determining web page texts users interested in according to browsed web pages | |
| CN112380431B (en) | Big data-based intelligent tourism analysis decision system and method | |
| CN103473291A (en) | Personalized service recommendation system and method based on latent semantic probability models | |
| KR20140071417A (en) | Mobile advertising using data networks based on intelligence data associated with internet-connectable devices derived using graph models | |
| CN102611785A (en) | Personalized active news recommending service system and method for mobile phone user | |
| CN104935963A (en) | Video recommendation method based on timing sequence data mining | |
| KR101174213B1 (en) | System and method for providing search result based on personal network | |
| CN103106208A (en) | Streaming media content recommendation method and system in mobile internet | |
| JP2013058213A5 (en) | ||
| CN105205689A (en) | Method and system for recommending commercial tenant | |
| JP6947191B2 (en) | Information processing equipment, information processing methods, and programs | |
| KR102020551B1 (en) | Method and Apparatus for Recommending Restaurant Based on Mobile Context Information | |
| CN106682152A (en) | Recommendation method for personalized information | |
| CN106375369A (en) | Mobile Web service recommendation method and collaborative recommendation system based on user behavior analysis | |
| CN103020066A (en) | Method and device for recognizing search demand | |
| CN107657347A (en) | Method is recommended in a kind of tourism towards group | |
| CN106980646A (en) | The method applied based on popularity to the influencing mechanism analysis of user interest and its in proposed algorithm | |
| CN108446351A (en) | The hotel's screening technique and system based on user preference of OTA platforms |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 100070, No. 101-8, building 1, 31, zone 188, South Fourth Ring Road, Beijing, Fengtai District Applicant after: Guoxin Youyi Data Co., Ltd Address before: 100070, No. 188, building 31, headquarters square, South Fourth Ring Road West, Fengtai District, Beijing Applicant before: SIC YOUE DATA Co.,Ltd. |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180209 |