CN109558983A - Network courses dropping rate prediction technique and device - Google Patents
Network courses dropping rate prediction technique and device Download PDFInfo
- Publication number
- CN109558983A CN109558983A CN201811467522.7A CN201811467522A CN109558983A CN 109558983 A CN109558983 A CN 109558983A CN 201811467522 A CN201811467522 A CN 201811467522A CN 109558983 A CN109558983 A CN 109558983A
- Authority
- CN
- China
- Prior art keywords
- matrix
- behavior
- dropping rate
- rate prediction
- variety
- 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
- 238000000034 method Methods 0.000 title claims abstract description 36
- 239000011159 matrix material Substances 0.000 claims abstract description 200
- 239000013598 vector Substances 0.000 claims abstract description 98
- 230000003542 behavioural effect Effects 0.000 claims abstract description 78
- 241001269238 Data Species 0.000 claims abstract description 19
- 238000012360 testing method Methods 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 25
- 238000012795 verification Methods 0.000 claims description 16
- 238000004422 calculation algorithm Methods 0.000 claims description 15
- 238000007689 inspection Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 abstract description 8
- 230000006399 behavior Effects 0.000 description 126
- 238000010586 diagram Methods 0.000 description 14
- 239000000047 product Substances 0.000 description 8
- 230000006870 function Effects 0.000 description 7
- 238000013480 data collection Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 239000006227 byproduct Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007635 classification algorithm Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 230000036461 convulsion Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008439 repair process Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
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
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
- G06Q50/205—Education administration or guidance
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of network courses dropping rate prediction technique and device, it is related to technical field of data processing, method includes obtaining multiple sample datas, and each sample data includes the dropping rate of user, a variety of behavior property information and the corresponding behavioral data of every kind of behavior property information, based on a variety of behavior property information architecture behavioural matrixes, and it based on the behavioural matrix carries out that the corresponding weight vectors of behavior matrix are calculated, a dropping rate prediction model is obtained according to the corresponding behavioral data of a variety of behavior property information of the dropping rate of each sample data and user and the weight vectors, and it is handled to obtain the network courses dropping rate prediction result of the learner based on a variety of behavioural informations of the dropping rate prediction model to learner.By the above method, so that dropping rate prediction model has the higher degree of convergence and accuracy, and then effectively avoid carrying out existing accuracy low problem when the prediction of network courses dropping rate.
Description
Technical field
The present invention relates to technical field of data processing, in particular to a kind of network courses dropping rate prediction technique and
Device.
Background technique
With the fast development of internet and education big data, online course is more more and more universal, all occurs both at home and abroad
A large amount of Network Learning Platform, enrollment has all reached up to a million, and platform enrollment can be more and more, but evidence
Investigation statistics find course completion rate it is generally low, therefore, for Network Learning Platform network courses dropping rate prediction at
For current research hotspot.
Data such as webpage click, video-see, operation submission, forum's behavior and examination etc. are generallyd use at present as user
Network courses dropping rate prediction algorithm, inventor it has been investigated that, face big data environment when, in face of different network sciences
Practise platform, the harmonies of data volume and data may biggish gap, and then network courses dropping rate prediction result can be made
Inaccuracy.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of network courses dropping rate prediction technique and devices, with effective
Alleviate above-mentioned technical problem.
To achieve the above object, the embodiment of the present invention adopts the following technical scheme that
A kind of network courses dropping rate prediction technique, comprising:
Obtain multiple sample datas, wherein each sample data include the dropping rate of user, a variety of behavior property information with
And the corresponding behavioral data of every kind of behavior property information；
Based on a variety of behavior property information architecture behavioural matrixes, and based on the behavioural matrix carry out that this is calculated
The corresponding weight vectors of behavioural matrix；
The weight vectors are verified, and after verification passes through, according to the dropping rate and use of each sample data
The corresponding behavioral data of a variety of behavior property information at family and the weight vectors use default sorting algorithm to be handled with
Obtain a dropping rate prediction model；
A variety of behavioural informations of learner are obtained, and a variety of behavioural informations are carried out using the dropping rate prediction model
Processing obtains the network courses dropping rate prediction result of the learner.
Optionally, in above-mentioned network courses dropping rate prediction technique, it is based on the behavior property information architecture behavior square
Battle array, and based on the behavioural matrix carry out that the corresponding weight vectors step of each behavior property information is calculated include:
A variety of behavior property information are divided to obtain multiple classification according to study scene, and according to each described
Corresponding behavior property information of classifying obtains a hierarchical model, wherein each corresponding at least one behavioural information of the classification；
Behavior classification matrix is constructed according to the hierarchical model and corresponding behavior property matrix of respectively classifying, to the behavior
Attribute matrix and each behavior classification matrix are respectively calculated to obtain classified weight vector corresponding with the behavior classification matrix,
And behavior weight vectors corresponding with the behavior property matrix.
Optionally, in above-mentioned network courses dropping rate prediction technique, classify to the behavior property matrix and each behavior
Matrix is respectively calculated to obtain classified weight vector corresponding with the behavior classification matrix, and with the behavior property square
Battle array corresponding behavior weight vectors the step of include:
For each matrix in the behavior classification matrix and each behavior property matrix, by each row of the matrix
Weight obtains the first column vector that the corresponding weight total value of each row is constituted after carrying out tired multiply, according to the order of matrix number to first row
Each weight total value in vector carries out evolution, and treated that weight total value is normalized to obtain the square to each evolution
The corresponding weight vectors of battle array.
Optionally, in above-mentioned network courses dropping rate prediction technique, the step of testing to the weight vectors packet
It includes:
For each matrix in the behavior classification matrix and each behavior property matrix, by the corresponding power of the matrix
Weight vector is multiplied to obtain the corresponding product of each row respectively with every a line of the matrix, then by the corresponding product point of every a line
The value of row not corresponding with the weight vectors is divided by obtain the corresponding characteristic root of each row in the matrix；
By in the corresponding characteristic root of the matrix rows Maximum characteristic root and the data subtracted each other of the order of matrix number with
The data that the order subtracts one and obtains are divided by obtain the matrix coincident indicator, and search and the coincident indicator from preset table
Corresponding reference index, and the consistency ration being divided by the coincident indicator and the reference index is less than a preset value
When, the weight vectors verification passes through.
Optionally, in above-mentioned network courses dropping rate prediction technique, according to the dropping rate and use of each sample data
The corresponding behavioral data of a variety of behavior property information at family and the weight vectors use default sorting algorithm to be handled with
The step of obtaining a dropping rate prediction model include:
The multiple sample data point to training sample set and test sample are concentrated, wherein the training sample set and
The test sample concentration respectively includes multiple sample datas；
Dropping rate, a variety of behavior numbers for including to each sample data that the training sample is concentrated by svm classifier algorithm
It is believed that corresponding behavioral data is ceased and the weight vectors are trained, to obtain an initial model；
Each sample data that the test sample is concentrated is tested by the initial model, to obtain a dropping rate
Prediction model.
The present invention also provides a kind of network courses dropping rate prediction meanss, comprising:
Sample acquisition module, for obtaining multiple sample datas, wherein each sample data include user dropping rate,
A variety of behavior property information and the corresponding behavioral data of every kind of behavior property information；
Matrix obtains module, for being based on a variety of behavior property information architecture behavioural matrixes, and is based on the behavior
Matrix carries out that the corresponding weight vectors of behavior matrix are calculated；
Model obtains module, for verifying to the weight vectors, and after verification passes through, according to each sample
The corresponding behavioral data of a variety of behavior property information of the dropping rate and user of data and the weight vectors are using default point
Class algorithm is handled to obtain a dropping rate prediction model；
Dropping rate prediction module for obtaining a variety of behavioural informations of learner, and uses the dropping rate prediction model
A variety of behavioural informations are handled to obtain the network courses dropping rate prediction result of the learner.
Optionally, in above-mentioned network courses dropping rate prediction meanss, the matrix obtains module and includes:
Classification submodule, for being divided to obtain multiple points according to study scene by a variety of behavior property information
Class, and a hierarchical model is obtained according to each corresponding behavior property information of classifying, wherein each classification is corresponding extremely
A kind of few behavioural information；
Weight obtains submodule, for constructing behavior classification matrix according to the hierarchical model and corresponding behavior of respectively classifying
Attribute matrix is respectively calculated to obtain and the behavior classification matrix to the behavior property matrix and each behavior classification matrix
Corresponding classified weight vector, and behavior weight vectors corresponding with the behavior property matrix.
Optionally, in above-mentioned network courses dropping rate prediction meanss, the weight obtains submodule, is also used to for institute
Each matrix in behavior classification matrix and each behavior property matrix is stated, after the weight of each row of the matrix is carried out tired multiply
The first column vector that the corresponding weight total value of each row is constituted is obtained, according to the order in the matrix to each of first column vector
Weight total value carries out evolution, and treated that weight total value is normalized to obtain the corresponding weight of the matrix to each evolution
Vector.
Optionally, in above-mentioned network courses dropping rate prediction meanss, the model obtains module and includes:
Characteristic root obtains submodule, for being directed to each of the behavior classification matrix and each behavior property matrix
The corresponding weight vectors of the matrix are multiplied to obtain the corresponding product of each row with every a line of the matrix, so by matrix respectively
The corresponding product of every a line is divided by to obtain afterwards with the value of the weight vectors respective column the corresponding feature of each row in the matrix respectively
Root；
Verify submodule, for by the corresponding characteristic root of the matrix rows Maximum characteristic root and the order of matrix number phase
The data that the data and the order subtracted subtract one and obtain are divided by obtain the matrix coincident indicator, and search from preset table
Reference index corresponding with the coincident indicator, and the consistency ration being divided by the coincident indicator and the reference index
When less than a preset value, the weight vectors verification passes through.
Optionally, in above-mentioned network courses dropping rate prediction meanss, the model obtains module further include:
Submodule is divided, for concentrating the multiple sample data point to training sample set and test sample, wherein institute
It states training sample set and test sample concentration respectively includes multiple sample datas；
Training submodule, for being stopped by svm classifier algorithm to what each sample data that the training sample is concentrated included
Rate, the corresponding behavioral data of a variety of behavioral data information and the weight vectors are trained, initial to obtain one
Model；
Submodule is tested, for surveying by the initial model to each sample data that the test sample is concentrated
Examination, to obtain a dropping rate prediction model.
A kind of network courses dropping rate prediction technique provided by the invention and device, pass through a variety of behavior property information architectures
Behavioural matrix, and carry out that the corresponding weight vectors of behavior matrix are calculated based on the behavioural matrix, and be based on the weight
The corresponding behavioral data of a variety of behavior property information of vector and dropping rate and user in each sample data is using default classification
Algorithm is handled to obtain dropping rate prediction model, and then keeps the degree of convergence of the dropping rate prediction model high and accuracy height, into
And predict the network courses dropping rate for being predicted to obtain the learner to a variety of behavioural informations of learner based on this model
As a result more accurate.
To enable the above objects, features and advantages of the present invention to be clearer and more comprehensible, preferred embodiment is cited below particularly, and cooperate
Appended attached drawing, is described in detail below.
Detailed description of the invention
Fig. 1 is the connection block diagram of electronic equipment provided in an embodiment of the present invention.
Fig. 2 is the flow diagram of network courses dropping rate prediction technique provided in an embodiment of the present invention.
Fig. 3 is the flow diagram of step S120 in Fig. 2.
Fig. 4 is the flow diagram of step S130 in Fig. 2.
Fig. 5 is another flow diagram of step S130 in Fig. 2.
Fig. 6 is a kind of schematic diagram of hierarchical model provided in an embodiment of the present invention.
Fig. 7 is a kind of connection block diagram of network courses dropping rate prediction meanss provided in an embodiment of the present invention.
Fig. 8 is the connection block diagram that a kind of matrix provided in an embodiment of the present invention obtains module.
Fig. 9 is the connection block diagram that a kind of model provided in an embodiment of the present invention obtains module.
Icon: 10- electronic equipment；12- memory；14- processor；100- network courses dropping rate prediction meanss；110-
Sample acquisition module；120- matrix obtains module；122- classification submodule；124- weight obtains submodule；130- model obtains
Module；132- characteristic root obtains submodule；134- verifies submodule；135- divides submodule；136- trains submodule；137- is surveyed
Swab module；140- dropping rate prediction module.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment only
It is a part of the embodiments of the present invention, instead of all the embodiments.The present invention being usually described and illustrated herein in the accompanying drawings
The component of embodiment can be arranged and be designed with a variety of different configurations.
Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimed
The scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiments of the present invention, this field is common
Technical staff's every other embodiment obtained without creative efforts belongs to the model that the present invention protects
It encloses.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, does not then need that it is further defined and explained in subsequent attached drawing.
As shown in Figure 1, the embodiment of the invention provides a kind of electronic equipment 10, including memory 12 and processor 14, institute
It states and is directly or indirectly electrically connected between memory 12 and the processor 14, to realize the transmission or interaction of data.
The storage is stored in specifically, being stored in the memory 12 in the form of software or firmware (Firmware)
Software function module in device 12, the software program and module that the processor 14 is stored in memory 12 by operation,
Such as the network courses dropping rate prediction meanss 100 in the embodiment of the present invention, thereby executing various function application and data processing,
Realize the network courses dropping rate prediction technique in the embodiment of the present invention.
The memory 12 may be, but not limited to, random access memory 12 (Random Access Memory,
RAM), read-only memory 12 (Read Only Memory, ROM), (the Programmable Read- of programmable read only memory 12
Only Memory, PROM), erasable read-only memory 12 (Erasable Programmable Read-Only Memory,
EPROM), (the Electrically Erasable Programmable Read-Only of electricallyerasable ROM (EEROM) 12
Memory, EEPROM) etc..Wherein, memory 12 is for storing program, and the processor 14 is held after receiving and executing instruction
Row described program.
The processor 14 can be general processor 14, including (the Central Processing of central processing unit 14
Unit, CPU), network processing unit 14 (Network Processor, NP) etc., can also be digital signal processor 14 (DSP),
Specific integrated circuit (ASIC), field programmable gate array (FPGA) either other programmable logic device, discrete gate or crystalline substance
Body pipe logical device, discrete hardware components.May be implemented or execute disclosed each method in the embodiment of the present invention, step and
Logic diagram.General processor 14 can be microprocessor 14 or the processor 14 is also possible to any conventional processor 14
Deng.
It is appreciated that structure shown in FIG. 1 is only to illustrate, the electronic equipment 10 may also include more than shown in Fig. 1
Perhaps less component or with the configuration different from shown in Fig. 1.Each component shown in Fig. 1 can use hardware, software
Or combinations thereof realize.
Incorporated by reference to Fig. 2, the present invention provides a kind of network courses dropping rate prediction side that can be applied to above-mentioned electronic equipment 10
Method, the network courses dropping rate prediction technique are applied to realize tetra- steps of step S110-S140 when above-mentioned electronic equipment 10.
Step S110: multiple sample datas are obtained, wherein each sample data includes the dropping rate of user, a variety of behaviors
Attribute information and the corresponding behavioral data of every kind of behavior property information.
It is appreciated that same user may carry out simultaneously the study of multiple network courses in Network Learning Platform, then it is same
A sample data is corresponding with when user learns each course respectively, i.e., the corresponding network courses of each sample data
It can be identical, be also possible to different.
The mode for obtaining multiple sample datas can be the multiple sample datas for obtaining user mode input, be also possible to obtain
The sample data stored in Network Learning Platform is taken, is not specifically limited herein, is configured according to actual needs.
For convenient for obtaining the sample data, and ensure the accuracy of the sample data obtained, in the present embodiment, institute
Stating step S110 may is that the sample data stored in the database for obtaining Network Learning Platform.
The mode for obtaining the sample data stored in the database of Network Learning Platform can be from database largely
It is extracted in user behavior event and event relation data, statistics further is carried out to data using data gathering tool and is obtained.
It should be noted that the type for the behavior property information for including in each sample data is identical, specifically, described
A variety of behavior property information in sample data are to carry out for consolidated network course, and a variety of behavior properties and correspondence
Behavioral data may include but being not limited to: user's course web-page requests and corresponding user carry out the number of web-page requests, use
The work of the quantity, user of the online hours of the online situation at family and user, the page situation of user's access and user to access pages
Jerk and user enliven number of days, user video course plays the video-see of the video playing number of situation and user, user
Situation and the video-see duration of user, the video-see quantity of user video viewing total quantity and user, user's viewing view
Frequency pause conditions and user watch the pause frequency, Switch Video broadcasting speed situation and the user's Switch Video broadcasting speed of video
Number, video playing terminates situation and video-see terminates number, attempts to do the homework and situation and attempt do the homework number, answer
Situation and answer quantity, Inspection situation and Inspection number carry out display answer situation and display answer number etc..
Step S120: a variety of behavior property information architecture behavioural matrixes are based on, and are carried out based on the behavioural matrix
The corresponding weight vectors of behavior matrix are calculated.
Step S130: verifying the weight vectors, and after verification passes through, according to stopping for each sample data
The corresponding behavioral data of a variety of behavior property information of rate and user and the weight vectors using default sorting algorithm into
Row processing is to obtain a dropping rate prediction model.
Step S140: a variety of behavioural informations of learner are obtained, and using the dropping rate prediction model to a variety of rows
It is handled to obtain the network courses dropping rate prediction result of the learner for information.
By the above method, so that using preset algorithm and based on multiple sample datas and according to the row in sample data
There is the higher degree of convergence and accuracy for the dropping rate prediction model that the corresponding weight vectors of attribute information obtain, and then effectively
Avoid carrying out existing accuracy low problem when the prediction of network courses dropping rate.
Specifically, step S120, which can be, constructs a behavioural matrix according to a variety of behavior properties, it is each in behavior matrix
The value of element can be policymaker and be compared according to the corresponding index of every two behavior property information, determine the ratio of the two
Example is determined with digital 1-n and its inverse.Specifically, using aijIndicate i-th of index to the comparison result of j-th of index then
Centainly haveAnd the corresponding weight of each behavior property information is obtained to obtain behavior square based on behavior matrix
The corresponding weight vectors of battle array.
Step S120 can also be that policymaker classifies according to by the study scene of a variety of behavior property information, such as can
Classified with clicking behavior according to webpage click behavior, video click behavior and examination question, obtains classification matrix and each point
The corresponding behavioural matrix of class, and be directed to each matrix, in the matrix every two classification or the corresponding finger of behavior property information
Mark is compared, and is determined the ratio of the two, matrix is determined with digital 1-n and its inverse, specifically, using aijIndicate i-th of finger
It marks to the comparison result of j-th of index then resultAnd the corresponding power of each behavior property information is obtained based on the matrix
Weight is to obtain the corresponding weight vectors of the matrix.
Incorporated by reference to Fig. 3, optionally, in the present embodiment, step S120 includes:
Step S122: a variety of behavior property information are divided to obtain multiple classification, and root according to study scene
A hierarchical model is obtained according to each corresponding behavior property information of classifying, wherein each classification is corresponding at least one
Behavioural information.
It according to study scene partitioning is webpage click by a variety of behavior property information specifically, in the present embodiment
Behavior is clicked in behavior, video and examination question clicks behavior, to obtain the corresponding behavior property information of classification of different study scenes,
To obtain the tree-like hierarchy model of three-decker.It is appreciated that carry out partition process in, can also for each study scene into
Row is further to be divided, to obtain the tree-like hierarchy model of multilayered structure.
Step S124: constructing behavior classification matrix according to the hierarchical model and corresponding behavior property matrix of respectively classifying,
The behavior property matrix and each behavior classification matrix are respectively calculated to obtain and corresponding point of the behavior classification matrix
Class weight vectors, and behavior weight vectors corresponding with the behavior property matrix.
Specifically, constructing the side of behavior classification matrix and the corresponding behavior property matrix of each classification by the hierarchical model
Formula is specially to be directed to each matrix, classifies to the every two in the matrix or the corresponding index of behavior property information is compared,
The ratio for determining the two, determines matrix with digital 1-n and its inverse, specifically, using aijIndicate that i-th of index refers to j-th
Target comparison result then resultAnd each classification or the corresponding weight of behavior property information are obtained based on the matrix, from
And obtain the corresponding weight vectors of the matrix.
Specifically, in the present embodiment, the mode for obtaining the corresponding weight vectors of each matrix may is that for the behavior
Each matrix in classification matrix and each behavior property matrix obtains respectively after the weight of each row of the matrix is carried out tired multiply
The first column vector that the corresponding weight total value of row is constituted, according to the order of matrix number to each weight total value in the first column vector
Evolution is carried out, and treated that weight total value is normalized to obtain the corresponding weight vectors of the matrix to each evolution.
By the above method, so that the weighted value obtained is more acurrate, and then when carrying out the prediction of network courses dropping rate, make
It is more accurate to obtain network courses dropping rate prediction result.
Network courses dropping rate prediction result further to ensure acquisition is more accurate, incorporated by reference to Fig. 4, in above-mentioned steps
The mode verified in S130 to the weight vectors can specifically include:
Step S132: for each matrix in the behavior classification matrix and each behavior property matrix, by the square
The corresponding weight vectors of battle array are multiplied to obtain the corresponding product of each row respectively with every a line of the matrix, then by every a line pair
The product answered is divided by obtain respectively with the value of the weight vectors respective column the corresponding characteristic root of each row in the matrix.
Specifically, the corresponding weight vectors of the matrix are multiplied to obtain a product with the first row of the matrix, and this is multiplied
It is long-pending to be divided by obtain the corresponding characteristic root of the first row of the matrix with the value of the first row of the weight vectors, using same as described above
The corresponding characteristic root of other rows in the matrix can be obtained in calculating process.
Step S134: the Maximum characteristic root in the corresponding characteristic root of the matrix rows is subtracted each other to obtain with the order of matrix number
Data and the order data that subtract one and obtain be divided by obtain the matrix coincident indicator, and search from preset table with this one
The cause property corresponding reference index of index, and the consistency ration being divided by the coincident indicator and the reference index is less than one
When preset value, the weight vectors verification passes through.
Wherein, multiple coincident indicator and the corresponding reference index of each coincident indicator are prestored in the preset table.
The preset value can be 0.09,0.1,0.11 or 0.12, be not specifically limited herein, and be configured according to actual needs i.e.
It can.
For further make obtain dropping rate prediction model the degree of convergence it is higher, and make obtain prediction result it is more quasi-
Really, incorporated by reference to Fig. 5, in the present embodiment, the step S130 further include:
Step S135: the multiple sample data point to training sample set and test sample are concentrated, wherein the training
Sample set and test sample concentration respectively include multiple sample datas.
Step S136: the dropping rate that includes to each sample data that the training sample is concentrated by svm classifier algorithm, more
The kind corresponding behavioral data of behavioral data information and the weight vectors are trained, to obtain an initial model.
Step S137: testing each sample data that the test sample is concentrated by the initial model, with
To a dropping rate prediction model.
Wherein, to keep the disaggregated model of above-mentioned acquisition more accurate, multiple target sample data divide
To training sample data collection and test sample data set in, training sample data concentrate include target sample data quantity
It may be, but not limited to, 7:3 or 8:2 with the ratio of the quantity for the target sample data for including in the test sample data set
Deng being not specifically limited herein.
In the present embodiment, by multiple behavior property information in the sample data be divided into webpage click behavior,
Video clicks behavior and examination question clicks behavior and classifies, obtained sorted hierarchical model such as Fig. 6.
The classification matrix constructed according to above-mentioned hierarchical model and corresponding behavior property matrix of respectively classifying are as follows:
1 classification matrix A of table
2 webpage click behavioural matrix B1 of table
3 video of table clicks behavioural matrix B2
4 examination question of table clicks behavioural matrix B3
It is specific in such a way that the hierarchical model constructs behavior classification matrix and the corresponding behavior property matrix of each classification
To classify to the every two in the matrix or the corresponding index of behavior property information being compared, determine two for each matrix
The ratio of person determines matrix with digital 1-n and its inverse, specifically, using aijIndicate i-th of index to the ratio of j-th of index
Compared with result then resultAnd each classification or the corresponding weight of behavior property information are obtained to obtain based on the matrix
The corresponding weight vectors of the matrix.
It, will be in each weight vectors and each sample data after being verified to the corresponding weight vectors of each matrix of above-mentioned acquisition
Dropping rate and the corresponding behavioral data of each behavior property information be trained and verification obtains the dropping rate using SVM algorithm
Prediction model, so that model accuracy with higher and convergence rate, avoid in the prior art in face of unbalanced number
According to when collection or when big-sample data collection, existing convergence rate is low, and the problem that training effectiveness is low.Therefore, the application exploitation right
Weight matrix can preferably improve the convergence rate and accuracy of model, provide for large sample and the training of unbalanced data set
New thinking.
Referring to Fig. 7, on the basis of the above, the present invention also provides a kind of network courses dropping rate prediction meanss 100, including
Sample acquisition module 110, matrix obtain module 120, model obtains module 130 and dropping rate prediction module 140.
The sample acquisition module 110, for obtaining multiple sample datas, wherein each sample data includes user's
Dropping rate, a variety of behavior property information and the corresponding behavioral data of every kind of behavior property information.In the present embodiment, the sample
This acquisition module 110 can be used for executing step S110 shown in Fig. 2, and the specific descriptions about the sample acquisition module 110 can
Referring to the description to step S110 above.
The matrix obtains module 120, for being based on a variety of behavior property information architecture behavioural matrixes, and is based on institute
Behavioural matrix is stated to carry out that the corresponding weight vectors of behavior matrix are calculated.In the present embodiment, the matrix obtains module
120 can be used for executing step S120 shown in Fig. 2, and the specific descriptions for obtaining module 120 about the matrix are referred to above
Description to step S120.
In the present embodiment incorporated by reference to Fig. 8, it includes that classification submodule 122 and weight obtain that the matrix, which obtains module 120,
Submodule 124.
The classification submodule 122, for being divided to obtain according to study scene by a variety of behavior property information
Multiple classification, and a hierarchical model is obtained according to each corresponding behavior property information of classifying, wherein each classification
Corresponding at least one behavioural information.In the present embodiment, the classification submodule 122 can be used for executing step shown in Fig. 3
S122, the specific descriptions about the classification submodule 122 are referred to the description to step S122 above.
The weight obtains submodule 124, for constructing behavior classification matrix and each classification pair according to the hierarchical model
The behavior property matrix answered is respectively calculated to obtain and the behavior to the behavior property matrix and each behavior classification matrix
The corresponding classified weight vector of classification matrix, and behavior weight vectors corresponding with the behavior property matrix.In this implementation
In example, the weight, which obtains submodule 124, can be used for executing step S124 shown in Fig. 3, obtain submodule about the weight
124 specific descriptions are referred to the description to step S124 above.
Wherein, the weight obtains submodule 124, is also used to for the behavior classification matrix and each behavior property
The weight of each row of the matrix, obtain after tired multiply that the corresponding weight total value of each row constitutes the by each matrix in matrix
One column vector, according to the order in the matrix in the first column vector each weight total value carry out evolution, and to each evolution at
Weight total value after reason is normalized to obtain the corresponding weight vectors of the matrix.
The model obtains module 130, for verifying to the weight vectors, and after verification passes through, according to each
The corresponding behavioral data of a variety of behavior property information of the dropping rate and user of the sample data and the weight vectors are adopted
It is handled with default sorting algorithm to obtain a dropping rate prediction model.In the present embodiment, the model obtains module 130
It can be used for executing step S130 shown in Fig. 2, the specific descriptions for obtaining module 130 about the model are referred to above to step
The description of rapid S130.
Referring to Fig. 9, in the present embodiment, it includes that characteristic root obtains submodule 132 and school that the model, which obtains module 130,
Test submodule 134.
The characteristic root obtains submodule 132, for being directed to the behavior classification matrix and each behavior property matrix
In each matrix, the corresponding weight vectors of the matrix are multiplied respectively obtain each row with every a line of the matrix corresponding
Then the corresponding product of every a line is divided by obtain each row in the matrix with the value of the weight vectors respective column corresponding by product respectively
Characteristic root.In the present embodiment, the characteristic root, which obtains submodule 132, can be used for executing step S132 shown in Fig. 4, about
The specific descriptions that the characteristic root obtains submodule 132 are referred to the description to step S132 above.
The verification submodule 134, for by the corresponding characteristic root of the matrix rows Maximum characteristic root and the matrix
The data subtracted each other of order and the order data that subtract one and obtain be divided by obtain the matrix coincident indicator, and from preset table
Reference index corresponding with the coincident indicator, and one to be divided by the coincident indicator and the reference index are searched in lattice
When causing sex ratio less than a preset value, the weight vectors verification passes through.In the present embodiment, the verification submodule 134 can
For executing step S134 shown in Fig. 4, the specific descriptions about the verification submodule 134 are referred to above to step
The description of S134.
In the present embodiment, the model obtain module 130 further include divide submodule 135, training submodule 136 and
Test submodule.
The division submodule 135, for the multiple sample data point to training sample set and test sample to be concentrated,
Wherein, the training sample set and test sample concentration respectively include multiple sample datas.In the present embodiment, described stroke
Molecular modules 135 can be used for executing step S135 shown in fig. 5, can join about the specific descriptions for dividing submodule 135
According to the description above to step S135.
The trained submodule 136, each sample data packet for being concentrated by svm classifier algorithm to the training sample
Dropping rate, the corresponding behavioral data of a variety of behavioral data information and the weight vectors included are trained, to obtain
One initial model.In the present embodiment, the trained submodule 136 can be used for executing step S136 shown in fig. 5, about described
The specific descriptions of training submodule 136 are referred to the description to step S136 above.
The test submodule 137, each sample data for being concentrated by the initial model to the test sample
It is tested, to obtain a dropping rate prediction model.In the present embodiment, the test submodule 137 can be used for executing Fig. 5 institute
The step S138 shown, the specific descriptions about the test submodule 137 are referred to the description to step S138 above.
The dropping rate prediction module 140, for obtaining a variety of behavioural informations of learner, and it is pre- using the dropping rate
Model is surveyed a variety of behavioural informations are handled to obtain the network courses dropping rate prediction result of the learner.In the present embodiment
In, the dropping rate prediction module 140 can be used for executing step S140 shown in Fig. 2, about the dropping rate prediction module 140
Specific descriptions be referred to the description to step S140 above.
To sum up, a kind of network courses dropping rate prediction technique provided by the invention and device, method include obtaining multiple samples
Notebook data, and each sample data includes dropping rate, a variety of behavior property information and the every kind of behavior property information pair of user
The behavioral data answered is based on a variety of behavior property information architecture behavioural matrixes, and is calculated based on the behavioural matrix
The corresponding weight vectors of behavior matrix are obtained, are believed according to the dropping rate of each sample data and a variety of behavior properties of user
It ceases corresponding behavioral data and the weight vectors obtains a dropping rate prediction model, and be based on the dropping rate prediction model pair
A variety of behavioural informations of learner are handled to obtain the network courses dropping rate prediction result of the learner.Pass through above-mentioned side
Method so that dropping rate prediction model has the higher degree of convergence and accuracy, and then effectively avoids carrying out network courses dropping rate
Existing accuracy low problem when prediction.
In several embodiments provided by the embodiment of the present invention, it should be understood that disclosed device and method, it can also
To realize by another way.Device and method embodiment described above is only schematical, for example, in attached drawing
Flow chart and block diagram show that the devices of multiple embodiments according to the present invention, method and computer program product are able to achieve
Architecture, function and operation.In this regard, each box in flowchart or block diagram can represent module, a program
A part of section or code, a part of the module, section or code include that one or more is patrolled for realizing defined
Collect the executable instruction of function.It should also be noted that in some implementations as replacement, function marked in the box
It can occur in a different order than that indicated in the drawings.For example, two continuous boxes can actually be held substantially in parallel
Row, they can also be executed in the opposite order sometimes, and this depends on the function involved.It is also noted that block diagram and/or
The combination of each box in flow chart and the box in block diagram and or flow chart, can the function as defined in executing or dynamic
The dedicated hardware based system made is realized, or can be realized using a combination of dedicated hardware and computer instructions.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of network courses dropping rate prediction technique characterized by comprising
Obtain multiple sample datas, wherein each sample data includes the dropping rate of user, a variety of behavior property information and every
The corresponding behavioral data of kind behavior property information；
Based on a variety of behavior property information architecture behavioural matrixes, and based on the behavioural matrix carry out that the behavior is calculated
The corresponding weight vectors of matrix；
The weight vectors are verified, and after verification passes through, according to the dropping rate of each sample data and user
A variety of corresponding behavioral datas of behavior property information and the weight vectors use default sorting algorithm to be handled to obtain
One dropping rate prediction model；
A variety of behavioural informations of learner are obtained, and a variety of behavioural informations are handled using the dropping rate prediction model
Obtain the network courses dropping rate prediction result of the learner.
2. network courses dropping rate prediction technique according to claim 1, which is characterized in that believed based on the behavior property
Breath building behavioural matrix, and based on the behavioural matrix carry out that the corresponding weight vectors of each behavior property information are calculated
Step includes:
A variety of behavior property information are divided to obtain multiple classification according to study scene, and according to each classification
Corresponding behavior property information obtains a hierarchical model, wherein each corresponding at least one behavioural information of the classification；
Behavior classification matrix is constructed according to the hierarchical model and corresponding behavior property matrix of respectively classifying, to the behavior property
Matrix and each behavior classification matrix are respectively calculated to obtain classified weight vector corresponding with the behavior classification matrix, and
Behavior weight vectors corresponding with the behavior property matrix.
3. network courses dropping rate prediction technique according to claim 2, which is characterized in that the behavior property matrix
Be respectively calculated to obtain classified weight vector corresponding with the behavior classification matrix with each behavior classification matrix, and with institute
The step of stating behavior property matrix corresponding behavior weight vectors include:
For each matrix in the behavior classification matrix and each behavior property matrix, by the weight of each row of the matrix
The first column vector that the corresponding weight total value of each row is constituted is obtained after carrying out tired multiply, according to the order of matrix number to the first column vector
In each weight total value carry out evolution, and treated that weight total value is normalized to obtain the matrix pair to each evolution
The weight vectors answered.
4. network courses dropping rate prediction technique according to claim 2, which is characterized in that carried out to the weight vectors
The step of inspection includes:
For each matrix in the behavior classification matrix and each behavior property matrix, by the corresponding weight of the matrix to
Amount be multiplied to obtain the corresponding product of each row respectively with every a line of the matrix, then by the corresponding product of every a line respectively with
The value that the weight vectors correspond to row is divided by obtain the corresponding characteristic root of each row in the matrix；
By in the corresponding characteristic root of the matrix rows Maximum characteristic root and the data subtracted each other of the order of matrix number and the rank
The data that number subtracts one and obtains are divided by obtain the matrix coincident indicator, and lookup is corresponding with the coincident indicator from preset table
Reference index, and when the consistency ration that the coincident indicator and the reference index are divided by is less than a preset value, institute
Weight vectors verification is stated to pass through.
5. network courses dropping rate prediction technique according to claim 1, which is characterized in that according to each sample data
Dropping rate and user the corresponding behavioral data of a variety of behavior property information and the weight vectors calculated using default classification
Method is handled to include: the step of obtaining a dropping rate prediction model
The multiple sample data point to training sample set and test sample are concentrated, wherein the training sample set and described
Test sample concentration respectively includes multiple sample datas；
Dropping rate, a variety of behavioral datas letter for including to each sample data that the training sample is concentrated by svm classifier algorithm
It ceases corresponding behavioral data and the weight vectors is trained, to obtain an initial model；
Each sample data that the test sample is concentrated is tested by the initial model, to obtain dropping rate prediction
Model.
6. a kind of network courses dropping rate prediction meanss characterized by comprising
Sample acquisition module, for obtaining multiple sample datas, wherein each sample data includes the dropping rate, a variety of of user
Behavior property information and the corresponding behavioral data of every kind of behavior property information；
Matrix obtains module, for being based on a variety of behavior property information architecture behavioural matrixes, and is based on the behavioural matrix
It carries out that the corresponding weight vectors of behavior matrix are calculated；
Model obtains module, for verifying to the weight vectors, and after verification passes through, according to each sample data
Dropping rate and user the corresponding behavioral data of a variety of behavior property information and the weight vectors calculated using default classification
Method is handled to obtain a dropping rate prediction model；
Dropping rate prediction module, for obtaining a variety of behavioural informations of learner, and using the dropping rate prediction model to this
A variety of behavioural informations are handled to obtain the network courses dropping rate prediction result of the learner.
7. network courses dropping rate prediction meanss according to claim 6, which is characterized in that the matrix obtains module packet
It includes:
Classification submodule, for a variety of behavior property information to be divided to obtain multiple classification according to study scene, and
A hierarchical model is obtained according to each corresponding behavior property information of classifying, wherein each classification corresponding at least one
Kind behavioural information；
Weight obtains submodule, for constructing behavior classification matrix according to the hierarchical model and corresponding behavior property of respectively classifying
Matrix is respectively calculated to obtain corresponding with the behavior classification matrix to the behavior property matrix and each behavior classification matrix
Classified weight vector, and behavior weight vectors corresponding with the behavior property matrix.
8. network courses dropping rate prediction meanss according to claim 7, which is characterized in that the weight obtains submodule
Block is also used to for each matrix in the behavior classification matrix and each behavior property matrix, by each row of the matrix
Weight carry out tired multiply after obtain the first column vector that the corresponding weight total value of each row is constituted, according to the order in the matrix to the
Each weight total value in one column vector carries out evolution, and treated that weight total value is normalized to obtain to each evolution
The corresponding weight vectors of the matrix.
9. network courses dropping rate prediction meanss according to claim 8, which is characterized in that the model obtains module packet
It includes:
Characteristic root obtains submodule, for for each square in the behavior classification matrix and each behavior property matrix
Battle array, the corresponding weight vectors of the matrix is multiplied respectively to obtain the corresponding product of each row with every a line of the matrix, then
The corresponding product of every a line is divided by to obtain with the value of the weight vectors respective column the corresponding characteristic root of each row in the matrix respectively；
Submodule is verified, for subtracting each other the Maximum characteristic root in the corresponding characteristic root of the matrix rows with the order of matrix number
To data and the order data that subtract one and obtain be divided by obtain the matrix coincident indicator, and searched from preset table and should
The corresponding reference index of coincident indicator, and be less than in the consistency ration that the coincident indicator is divided by with the reference index
When one preset value, the weight vectors verification passes through.
10. network courses dropping rate prediction meanss according to claim 6, which is characterized in that the model obtains module
Include:
Submodule is divided, for concentrating the multiple sample data point to training sample set and test sample, wherein the instruction
Practice sample set and test sample concentration respectively includes multiple sample datas；
Training submodule, dropping rate for including to each sample data that the training sample is concentrated by svm classifier algorithm,
A variety of corresponding behavioral datas of behavioral data information and the weight vectors are trained, to obtain an initial model；
Submodule is tested, for being tested by the initial model each sample data that the test sample is concentrated, with
Obtain a dropping rate prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811467522.7A CN109558983A (en) | 2018-12-03 | 2018-12-03 | Network courses dropping rate prediction technique and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811467522.7A CN109558983A (en) | 2018-12-03 | 2018-12-03 | Network courses dropping rate prediction technique and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109558983A true CN109558983A (en) | 2019-04-02 |
Family
ID=65868504
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811467522.7A Pending CN109558983A (en) | 2018-12-03 | 2018-12-03 | Network courses dropping rate prediction technique and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109558983A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110866162A (en) * | 2019-10-10 | 2020-03-06 | 西安交通大学 | Causal relationship mining method based on conjugate behaviors in MOOC data |
| CN112116137A (en) * | 2020-09-06 | 2020-12-22 | 桂林电子科技大学 | Student class dropping prediction method based on mixed deep neural network |
| CN112734105A (en) * | 2021-01-08 | 2021-04-30 | 浙江工业大学 | Method for preventing breaking behavior in online education |
| CN112862250A (en) * | 2021-01-12 | 2021-05-28 | 浙江知行教育科技有限公司 | College learning evaluation system and method based on big data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105007170A (en) * | 2015-05-11 | 2015-10-28 | 大连理工大学 | WLAN load comprehensive evaluation method based on FAHP-SVM theory |
| CN106776884A (en) * | 2016-11-30 | 2017-05-31 | 江苏大学 | A kind of act of terrorism Forecasting Methodology that multi-categorizer is combined based on multi-tag |
| CN107086922A (en) * | 2016-02-15 | 2017-08-22 | ***通信集团福建有限公司 | A kind of user behavior recognition method and apparatus |
| CN107231345A (en) * | 2017-05-03 | 2017-10-03 | 成都国腾实业集团有限公司 | Networks congestion control methods of risk assessment based on AHP |
| CN108121795A (en) * | 2017-12-20 | 2018-06-05 | 北京奇虎科技有限公司 | User's behavior prediction method and device |
-
2018
- 2018-12-03 CN CN201811467522.7A patent/CN109558983A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105007170A (en) * | 2015-05-11 | 2015-10-28 | 大连理工大学 | WLAN load comprehensive evaluation method based on FAHP-SVM theory |
| CN107086922A (en) * | 2016-02-15 | 2017-08-22 | ***通信集团福建有限公司 | A kind of user behavior recognition method and apparatus |
| CN106776884A (en) * | 2016-11-30 | 2017-05-31 | 江苏大学 | A kind of act of terrorism Forecasting Methodology that multi-categorizer is combined based on multi-tag |
| CN107231345A (en) * | 2017-05-03 | 2017-10-03 | 成都国腾实业集团有限公司 | Networks congestion control methods of risk assessment based on AHP |
| CN108121795A (en) * | 2017-12-20 | 2018-06-05 | 北京奇虎科技有限公司 | User's behavior prediction method and device |
Non-Patent Citations (3)
| Title |
|---|
| M. KLOFT 等: "Predicting MOOC Dropout over Weeks Using Machine Learning Methods", 《PROCEEDINGS OF THE 2014 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING》 * |
| 许艳秋，潘美芹: "层次分析法和支持向量机在个人信用评估中的应用", 《中国管理科学》 * |
| 郑石桥: "《管理审计方法》", 28 February 2017, 大连：东北财经大学出版社 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110866162A (en) * | 2019-10-10 | 2020-03-06 | 西安交通大学 | Causal relationship mining method based on conjugate behaviors in MOOC data |
| CN110866162B (en) * | 2019-10-10 | 2021-11-19 | 西安交通大学 | Causal relationship mining method based on conjugate behaviors in MOOC data |
| CN112116137A (en) * | 2020-09-06 | 2020-12-22 | 桂林电子科技大学 | Student class dropping prediction method based on mixed deep neural network |
| CN112734105A (en) * | 2021-01-08 | 2021-04-30 | 浙江工业大学 | Method for preventing breaking behavior in online education |
| CN112862250A (en) * | 2021-01-12 | 2021-05-28 | 浙江知行教育科技有限公司 | College learning evaluation system and method based on big data |
| CN112862250B (en) * | 2021-01-12 | 2023-12-26 | 北京漂洋过海科技有限责任公司 | College learning evaluation system and method based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109558983A (en) | Network courses dropping rate prediction technique and device | |
| CN105138653B (en) | It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty | |
| CN106201871A (en) | Based on the Software Defects Predict Methods that cost-sensitive is semi-supervised | |
| CN106547871A (en) | Method and apparatus is recalled based on the Search Results of neutral net | |
| Hu et al. | Identification of migratory insects from their physical features using a decision-tree support vector machine and its application to radar entomology | |
| CN107209853A (en) | Positioning and map constructing method | |
| CN105354595A (en) | Robust visual image classification method and system | |
| CN109597937A (en) | Network courses recommended method and device | |
| Peng | Assortative mixing, preferential attachment, and triadic closure: A longitudinal study of tie-generative mechanisms in journal citation networks | |
| CN105893390A (en) | Application program processing method and electronic equipment | |
| CN109840413A (en) | A kind of detection method for phishing site and device | |
| Caruana et al. | Mining citizen science data to predict orevalence of wild bird species | |
| Das et al. | An examination system automation using natural language processing | |
| Zhang et al. | An interpretable online learner's performance prediction model based on learning analytics | |
| CN113919510A (en) | Sample feature selection method, device, equipment and medium | |
| CN110262906B (en) | Interface label recommendation method and device, storage medium and electronic equipment | |
| CN104298997B (en) | data classification method and device | |
| CN110096708A (en) | A kind of determining method and device of calibration collection | |
| CN110427964A (en) | A kind of multivariate time series Variable Selection based on mutual information | |
| CN109377017A (en) | A kind of information system is practical and data health degree evaluation method | |
| CN113065975B (en) | Method, system and terminal for calculating focusing degree and evolution relation of network public sentiment topics | |
| CN105117385B (en) | A kind of method and system that public opinion information extraction is carried out based on matrix computations | |
| CN107845047A (en) | A kind of dynamic grading system, method and computer-readable recording medium | |
| CN105447018B (en) | Verify the method and device of Web page classifying model | |
| CN109213937A (en) | Intelligent search method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190402 |