CN109272164A - Learning behavior dynamic prediction method, device, equipment and storage medium - Google Patents
Learning behavior dynamic prediction method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN109272164A CN109272164A CN201811144725.2A CN201811144725A CN109272164A CN 109272164 A CN109272164 A CN 109272164A CN 201811144725 A CN201811144725 A CN 201811144725A CN 109272164 A CN109272164 A CN 109272164A
- Authority
- CN
- China
- Prior art keywords
- learning behavior
- user
- data
- learning
- dynamic prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 230000006399 behavior Effects 0.000 title claims abstract description 103
- 238000000034 method Methods 0.000 title claims abstract description 63
- 238000003860 storage Methods 0.000 title claims abstract description 10
- 238000013528 artificial neural network Methods 0.000 claims abstract description 23
- 230000006403 short-term memory Effects 0.000 claims abstract description 21
- 230000000306 recurrent effect Effects 0.000 claims abstract description 20
- 238000004458 analytical method Methods 0.000 claims abstract description 15
- 239000013598 vector Substances 0.000 claims description 33
- 238000010586 diagram Methods 0.000 claims description 21
- 230000006870 function Effects 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 17
- 238000012800 visualization Methods 0.000 claims description 13
- 230000015654 memory Effects 0.000 claims description 6
- 238000005457 optimization Methods 0.000 claims description 6
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 5
- 238000009826 distribution Methods 0.000 claims description 5
- 238000003062 neural network model Methods 0.000 claims description 5
- 230000003542 behavioural effect Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 4
- 230000003442 weekly effect Effects 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 239000000463 material Substances 0.000 claims description 3
- 230000000007 visual effect Effects 0.000 abstract description 9
- 238000007405 data analysis Methods 0.000 abstract description 6
- 230000008569 process Effects 0.000 abstract description 5
- 239000002131 composite material Substances 0.000 abstract description 2
- 230000000875 corresponding effect Effects 0.000 description 6
- 238000004422 calculation algorithm Methods 0.000 description 4
- 239000000725 suspension Substances 0.000 description 3
- 230000003044 adaptive effect Effects 0.000 description 2
- 230000003321 amplification Effects 0.000 description 2
- 238000013079 data visualisation Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000007787 long-term memory Effects 0.000 description 2
- 238000003199 nucleic acid amplification method Methods 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 241001275954 Cortinarius caperatus Species 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000013210 evaluation model Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 238000002360 preparation method Methods 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- Tourism & Hospitality (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Educational Technology (AREA)
- Biophysics (AREA)
- Educational Administration (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Life Sciences & Earth Sciences (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of learning behavior dynamic prediction method, device, equipment and storage mediums, the prediction technique comprising steps of based on user historical time section learning behavior data, it is trained using long short-term memory Recognition with Recurrent Neural Network as prediction model, obtains optimal models；The user is obtained in the learning behavior data of first time period, the optimal models is based on, exports prediction result.The present invention predicts output result as prediction model by using LSTM-RNN, the accuracy of prediction can be improved, and user information data are shown using interactivity visual analysis mode, it is convenient to check on the backstage of system, analyze user data information, to help teacher that can timely, intuitively know data analysis result, and then realize " closed loop " of teaching process.It the composite can be widely applied to various learning behavior forecasting systems.
Description
Technical field
The present invention relates to computer data analysis field more particularly to a kind of learning behavior dynamic prediction method, device, set
Standby and storage medium.
Background technique
The abbreviation of MOOC:massive open online courses, extensive open online course；
The abbreviation of LSTM:Long Short-Term Memory, long short-term memory；
The abbreviation of LSTM-RNN:Long Short-Term Memory-Recurrent Neural Networks, length
When remember Recognition with Recurrent Neural Network；
Adam algorithm: this name of Adam derive from adaptive moment estimation, adaptive moments estimation, this
It is a kind of algorithm for optimizing random targets function based on First-order Gradient.
Logistic Sigmoid function: value range is (0,1), and activation primitive, letter are commonly used as in deep learning
Counting expression formula is
The abbreviation of AUC:Area Under Curve, the area under ROC curve；AUC value is a probability value, is chosen when at random
A positive sample and a negative sample are selected, current sorting algorithm comes this positive sample according to the score value being calculated negative
Probability before sample is exactly AUC value；
TensorFlow: being that complicated data structure is transmitted in artificial intelligence nerve net to carry out analysis and treatment process
System；
Django:Django is the Web application framework an of open source code, is write as by Python.Using the frame of MVC
Frame mode, i.e. model M, view V and controller C；The frame is with Belgian Gypsy jazz guitarist Django
Reinhardt is named；
The abbreviation of SVG:Scalable Vector Graphics, refers to saleable vector graphics, is to describe two using XML
Tie up the language of figure and mapping program；
Random forest: random forest refers to setting a kind of classifier for being trained sample and predicting using more.
This special regression model of logic: it is a kind of disaggregated model, is indicated by conditional probability distribution, form is the logic of parametrization
This is carefully distributed.
From 2012, MOOC started worldwide to obtain extensive attention rate.With online education mould before
Formula is compared, and MOOC provides interaction platform for teacher and user, while increasing the links such as operation, examination, more close to tradition
The teaching pattern in classroom, therefore MOCC is very by the welcome of user.
In MOOC course, online teaching form and excessive student's quantity make teacher be difficult to collect the feedback of student.
MOOC platform often will record the learning behavior of user, collector journal data, by the analysis to these data, can predict user
Study can or can not be continued to participate in next week, to help MOOC to reach better teaching efficiency and improve the retention ratio of user, meanwhile,
Data analysis result can intervene the user that may move back class with assisted teacher or system, reduce the probability that user moves back class.
Hot spot in recent years had become to the research of MOOC data, there are the following problems in existing research:
1. prediction accuracy is not high；
2. lacking visual solution, it has not been convenient to be managed on the backstage MOCC.
Summary of the invention
The present invention is directed to solve at least some of the technical problems in related technologies.For this purpose, of the invention
One purpose is to provide a kind of learning behavior dynamic prediction method, can be improved prediction accuracy, and visually divide using interactivity
Analysis mode shows user information data.
The technical scheme adopted by the invention is that: a kind of learning behavior dynamic prediction method is provided, this method comprises: being based on
User is instructed using long short-term memory Recognition with Recurrent Neural Network as prediction model in the learning behavior data of historical time section
Practice, obtains optimal models；The user is obtained in the learning behavior data of first time period, is based on the optimal models, output prediction
As a result.
Further, it should be recycled based on user in the history learning behavioral data of historical time section using long short-term memory
The step of neural network is trained as prediction model, obtains optimal models includes: to obtain user in historical time section
Practise behavioral data；Learning characteristic vector [x is extracted based on the learning behavior data1,x2,..,xK] and label [y1,y2,..,yK],
Wherein, xiFor the corresponding learning characteristic vector of learning behavior data of user's i-th of unit time in the historical time section,
yiFor indicating whether user has learning behavior in the i+1 unit time, if any learning behavior, then yi=1, conversely, then yi=
0；Wherein, i=1,2,3K, K >=1；With learning characteristic vector [x1,x2,..,xK] it is input, with [y1,
y2,..,yK] it is label training pattern, obtain the optimal models.
It further, should be with learning characteristic vector [x1,x2,..,xK] it is input, [y1,y2,..,yK] it is label training mould
Type, is specifically included: being divided into multiple users according to the ratio of 4:1 using 5 folding cross-validation methods the step of obtaining the optimal models
Training set and verifying collection, and arameter optimization is carried out using trellis search method, obtain the optimal models.
When further, to long short-term memory Recognition with Recurrent Neural Network training, using intersection entropy function as loss function.
When further, to long short-term memory Recognition with Recurrent Neural Network training, parameter optimization is carried out using Adam method.
Further, which includes weight vectors and bias vector, which is [- 0.1,0.1]
It is uniformly distributed, which is null vector.
Further, the activation letter using Logistic Sigmoid function as the long short-term memory Recognition with Recurrent Neural Network
Number.
Further, this method further includes AUC verification step: selecting at least one set of baseline model；The baseline is respectively adopted
Model and the optimal models verify the verifying collection, obtain respective evaluation index AUC.
Further, this method further includes being statisticallyd analyze based on the learning behavior data and prediction result, obtains user's letter
Cease data；The user information data are shown using visualization system.
Further, visualization system is realized using B/S framework, using MySQL database storing data, using Django
Frame builds website, is drawn a diagram in html web page based on SVG element.
Further, visualization system based on the view that user information data are shown includes student information view, enlivens day
Go through view, studying progress arrangement view, personal footprint view.
Further, gender, age or the distribution of receiving an education of user are described in the student information view using cake chart.
Further, it is that daily any active ues number, the calendar are shown on a calendar figure that this, which enlivens calendar view,
Each rectangle in figure represents one day, and seven days in one week are placed in same row, and using course material update day as
First day of each column, the depth of rectangle color and any active ues number direct proportionality on the same day.
Further, the learning behavior data of each user are shown using three features in the studying progress arrangement view:
Operation number accounting, accumulative study trifle accounting, video playback time accounting, and using the prediction of the different color mark users
As a result.
Further, which, should by triggering graphic element pop-up in the studying progress arrangement view
The investment degree of user's study, the investment degree of user study are shown in personal footprint view in the form of stacking histogram
The number of video is being watched weekly and is submitting the number of operation including the user.
It is of the present invention another solution is that providing a kind of learning behavior dynamic prediction device, which includes
Neural network model generation module and learning behavior dynamic prediction module, the neural network model generation module are used to be based on user
In the learning behavior data of historical time section, it is trained, is obtained as prediction model using long short-term memory Recognition with Recurrent Neural Network
To optimal models；The learning behavior dynamic prediction module is for obtaining the user in the learning behavior data of first time period, base
In the optimal models, prediction result is exported.
Further, which further includes data statistic analysis model generation module and visible system generation module, the number
Analysis model generation module is used to statistically analyze based on the learning behavior data and prediction result according to statistics, obtains user information number
According to；The visible system generation module, for showing the user information data using visualization system.
Another technical solution of the present invention is: providing a kind of learning behavior Dynamic Forecasting System equipment, the equipment
It include: at least one processor；And the memory being connect at least one processor communication；Wherein, which stores
There is the instruction that can be executed by least one processor, which is executed by least one processor, so that this at least one
Processor is able to carry out above-mentioned method.
Yet another aspect of the present invention is: providing a kind of computer readable storage medium, this is computer-readable
Storage medium is stored with computer executable instructions, and the computer executable instructions are for making computer execute above-mentioned method.
The beneficial effects of the present invention are:
The present invention is trained using long short-term memory Recognition with Recurrent Neural Network as prediction model and tuning, obtains optimal mould
Type overcomes the not high technical problem of learning behavior prediction accuracy existing in the prior art, and it is accurate to realize a kind of prediction
Spend high learning behavior dynamic prediction method, device, equipment and storage medium.
In addition, the present invention also shows user information data by using interactivity visual analysis mode, it is convenient in system
Backstage is checked, analyzes user data information, to help teacher that can timely, intuitively know data analysis result, and then realizes
" closed loop " of teaching process.
It the composite can be widely applied to various learning behavior forecasting systems.
Detailed description of the invention
Fig. 1 is the flow diagram of an embodiment of learning behavior dynamic prediction method of the present invention；
Fig. 2 is the flow diagram of step S11 in Fig. 1；
Fig. 3 is the flow diagram of another embodiment of learning behavior dynamic prediction method of the present invention；
Fig. 4 is flow diagram of Fig. 3 embodiment under an application scenarios；
Fig. 5 is a kind of partial schematic diagram of embodiment of student information view；
Fig. 6 is to enliven a kind of partial schematic diagram of embodiment of calendar view；
Fig. 7 (a) and Fig. 7 (b) is a kind of partial schematic diagram of embodiment of studying progress arrangement view respectively；
Fig. 8 (a) and Fig. 8 (b) is a kind of partial schematic diagram of embodiment of personal footprint view respectively；
Fig. 9 is the structural schematic diagram of an embodiment of learning behavior dynamic prediction device of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, rather than whole embodiments.Based on this
Embodiment in invention, those of ordinary skill in the art are obtained every other under the premise of not making creative labor
Embodiment shall fall within the protection scope of the present invention.
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
Mutually combination.
Referring to Fig. 1, Fig. 1 is the flow diagram of an embodiment of learning behavior dynamic prediction method of the present invention.It should
Method includes:
Step S11: based on user historical time section learning behavior data, using long short-term memory Recognition with Recurrent Neural Network
It is trained as prediction model, obtains optimal models.
Wherein, historical time section can be last month, upper halves or previous year etc..The embodiment is with historical time section
For upper halves.
Learning behavior data include three classes data: activity log, lesson structure and personal information.For activity log, originally
Embodiment only considers event type relevant to learning behavior: video dependent event and exercise dependent event.The log sorted out
Event type is as shown in table 1:
1 logging time type of table
LSTM (shot and long term memory network) model specific algorithm is realized by three doors, and each is meant that control letter
The amount that breath is flowed into and reserved.Forgeing door indicates which historical data abandoned, and input gate control flows into the information of cell state, output
Which information outflow door determines.The information of previous moment, which reserves, simultaneously also flows into part for the information as subsequent time.Therefore
LSTM model may be implemented long-term memory, and information circulate wherein transmission when do not have nonlinear operation calculate thus can protect
It keeps steady surely constant.
Step S12: user is obtained in the learning behavior data of first time period, optimal models is based on, exports prediction result.
Wherein, first time period accordingly can be this month, this term or the current year.The embodiment is with first time period for this
For term.
In the present embodiment, using LSTM-RNN model as prediction model, prediction accuracy is high.With user's study class
When propulsion, step S12 iteration carry out, to realize dynamical output prediction result.
Referring to Fig.2, Fig. 2 is the flow diagram of step S11 in Fig. 1, step S11: based on user in historical time section
Learning behavior data are trained as prediction model using long short-term memory Recognition with Recurrent Neural Network, it is specific to obtain optimal models
Include:
S111: user is obtained in the learning behavior data of historical time section.
Wherein, the method for obtaining user in the learning behavior data of historical time section refers to preceding method.The step is gone through
For the history period above halves.
S112: learning characteristic vector [x is extracted based on the learning behavior data1,x2,..,xK] and label [y1,y2,..,
yK], wherein xiFor user's i-th of unit time in the historical time section the corresponding learning characteristic of learning behavior data to
Amount, yiFor indicating whether user has learning behavior in the i+1 unit time, if any learning behavior, then yi=1, conversely, then
yi=0；Wherein, i=1,2,3K, K >=1.
Wherein, using week as chronomere (1 week is 1 unit time), if the course period of term a branch of instruction in school is T
Week, user specific for one are drawn a portrait as feature vector with its i-th week learning behavior, are denoted as xi, then whole T weeks spies
Sign vector is denoted as [x1,x2,..,xT], whether there is learning behavior to be denoted as y i+1 weeki(if having learning behavior yi=1, without then
yi=0) whole T weeks label [y, are then obtained1,y2,..,yT]。
S113: with learning characteristic vector [x1,x2,..,xK] it is input, with [y1,y2,..,yK] it is label training pattern, it obtains
To the optimal models.
Wherein, with the propulsion of course, at the K weeks, by the end of the feature [x of current cycle1,x2,..,xK] it is defeated
Enter, [y1,y2,..,yK] it is label training pattern.
In the present embodiment, when to long short-term memory Recognition with Recurrent Neural Network training, loss is used as using entropy function is intersected
Function, and parameter optimization is carried out using Adam method, which includes weight vectors and bias vector.Using Logistic
Activation primitive of the Sigmoid function as the long short-term memory Recognition with Recurrent Neural Network, the value range of activation primitive be (0,
1), function expression isIn view of the function is close linear near 0, corresponding gradient is larger, power
The initial value of weight vector should be located near 0, therefore using being uniformly distributed to weight vectors and bigoted in [- 0.1,0.1] range
Vector is initialized.The hidden layer of long short-term memory Recognition with Recurrent Neural Network and output layer it is bigoted it is vector initialising be 0.
In this step, specifically, prediction technique is realized based on deep learning frame Tensorflow, in training, adopted
With 5 folding cross-validation methods by multiple users according to the ratio of 8:2 be divided into training set and verifying collect, and using trellis search method into
Row arameter optimization, obtains optimal models.
Preferably, this step further includes AUC verification step: selecting at least one set of baseline model；The baseline mould is respectively adopted
Type and the optimal models verify the verifying collection, obtain respective evaluation index AUC.
Since the prediction result of the present embodiment is two classification problems, the two common evaluation indexes of classification are selected
AUC carrys out the superiority and inferiority of evaluation model.The present embodiment has chosen two kinds of baseline models, respectively this special regression model (LR) of logic and with
Machine forest model (RF).As shown in table 2, the comparison of table 2 lists LSTM-RNN and two kinds of baseline models within the course period weekly
AUC score value.
| Week | LSTM-RNN | LR | RF |
| 1 | 0.52 | 0.50 | 0.51 |
| 2 | 0.54 | 0.52 | 0.51 |
| 3 | 0.55 | 0.51 | 0.52 |
| 4 | 0.60 | 0.58 | 0.57 |
| 5 | 0.66 | 0.62 | 0.64 |
| 6 | 0.71 | 0.62 | 0.67 |
| 7 | 0.73 | 0.65 | 0.66 |
| 8 | 0.82 | 0.66 | 0.73 |
| 9 | 0.84 | 0.71 | 0.72 |
| 10 | 0.87 | 0.75 | 0.78 |
| 11 | 0.85 | 0.70 | 0.79 |
| 12 | 0.81 | 0.68 | 0.73 |
| 13 | 0.84 | 0.72 | 0.76 |
| 14 | 0.83 | 0.69 | 0.74 |
2 LSTM-RNN model of table and baseline model AUC value
Can be found that from table 2: 1) the AUC score value highest of LSTM-RNN behaves oneself best.Since the 1st week, LSTM-RNN
Highest score is just achieved, and as course promotes, the advantage relative to baseline model is become readily apparent from；2) at first three
In week, for the AUC score value of three kinds of prediction techniques between 0.5~0.6, prediction result does not have reference value substantially；3) with cycle
Increase AUC score value to be gradually increased, this is because as data are constantly accumulated, it can be found that more learning behavior rules, are realized more
Add accurate prediction.But after the 10th week, AUC score value is just no longer improved, and is declined instead, this may will with course
Terminate related.
In the present embodiment, by with the learning characteristic vector [x of the unit time of user's history period1,x2,..,
xK] be input, unit time whether learnt [y1,y2,..,yK] it is label training pattern, obtain optimal models.And it is right
The superiority and inferiority of model is evaluated using AUC value, to prove that trained LSTM-RNN model is optimal models.
Referring to Fig. 3, Fig. 3 is the flow diagram of another embodiment of learning behavior dynamic prediction method of the present invention.
This method comprises:
Step S21: it based on user in the learning behavior data of historical time section (last term), is recycled using long short-term memory
Neural network is trained as prediction model, obtains optimal models.
Wherein, the implementation method of step S21 is with step S11, and therefore not to repeat here.
Step S22: user is obtained in the learning behavior data of first time period, optimal models is based on, exports prediction result.
Wherein, the implementation method of step S22 is with step S12, and therefore not to repeat here.
Step S23: user is obtained in the learning behavior data of first time period, optimal models is based on, exports prediction result.
Wherein, the implementation method of step S23 is with step S13, and therefore not to repeat here.
Step S24: it is statisticallyd analyze based on learning behavior data and prediction result, obtains user information data.
Wherein, user information data include student information view, enliven calendar view, studying progress arrangement view, individual
Footprint view.
Step S25: user information data are shown using visualization system.
Preferably, visualization system is realized using B/S framework, using MySQL database storing data, using Django frame
Frame builds website, is drawn a diagram in html web page based on SVG element.
Optionally, step S24 and step S25 can be interspersed in after step S21, step S22 or step S23 either step.
In the present embodiment, an interactivity visual analysis solution is devised, can show that historical data is united simultaneously
Result and prediction result are counted, system background is supplied to or teacher carries out auxiliary judgment.
Also referring to Fig. 3 to Fig. 8, as shown in figure 4, Fig. 4 is process signal of Fig. 3 embodiment under an application scenarios
Figure.As shown in figure 4, extracting the full course period learning behavior data of the upper halves a branch of instruction in school of user, it is assumed that the course
Whole periods are N weeks, carry out data prediction, obtain whole N weeks learning behavior feature vectors.At the K weeks, to learn spy
Levy vector [x1,x2,..,xK] it is input, with [y1,y2,..,yK] (K >=1) it is that label carries out model training and tuning, to obtain
Obtain optimal models.When course proceeds to this term, by the end of current date, it is assumed that course proceeds to i-th week, obtains user's sheet
The learning behavior data (available data) of the term course, carry out data prediction, and extraction obtains learning behavior vector [x1,
x2,..,xi] (learning behavior portrait), with learning behavior vector [x1,x2,..,xi] as input, [y1,y2,..,yi] as mark
Label are based on optimal models, to predict whether user learns (prediction result) in i+1 all (next week).As shown in figure 4, this is answered
It is also based on learning behavior data and prediction result statistical analysis with scene, obtains user information data, and use visualization system
Show user information data.Visualization system based on the view that user information data are shown includes student information view, enlivens day
Go through view, studying progress arrangement view, personal footprint view.Wherein, student information view can be generated according to learning behavior data,
Enlivening calendar view, studying progress arrangement view, personal footprint view can be according to learning behavior data and prediction result statistical
Analysis generates.
As shown in figure 5, Fig. 5 is the partial schematic diagram of Fig. 4 middle school student's information view.Student information view is for showing study
The personal background of person meets teaching team for the curiosity of " who is learning my course ".In the student information view, adopt
Gender, age or the distribution of receiving an education of user are described with cake chart.Fig. 5 show the student of " financial analysis and decision " course
Information view, it can be seen that having more than a student is male (Male), almost half student is undergraduate
(Bachelor), the students ' age of a quarter is had more than between 25 years old to 30 years old.
As shown in fig. 6, Fig. 6 is the partial schematic diagram for enlivening calendar view in Fig. 4.Enlivening calendar view is in a calendar
Daily any active ues number is shown on figure, each rectangle in the calendar figure represents one day, and seven days in one week are placed on
In same row (Thursday in Friday to next week is indicated with F, S, S, M, T, W, T respectively), and using course material update day as
First day of each column, the depth of rectangle color and any active ues number direct proportionality on the same day, color is deeper, shows to live
Jump number is more.Due to narrow space, suspension prompting frame is additionally provided to show specific number.When mouse is suspended in some square
When above shape, date on the same day, cycle, the number of enlivening can be shown in prompting frame, and color can be also emphasised in legend, with aobvious
That shows this day enlivens the relative level of number in history.Specifically as shown in fig. 6, being selected in legend by mobile two handles
Some range is selected, enlivening date of the number in range can be shown with corresponding color, and enliven number outside range of choice
Date can be shown as white.
As shown in Fig. 7 (a) and Fig. 7 (b), Fig. 7 (a) and Fig. 7 (b) is the part of studying progress arrangement view in Fig. 4 respectively
Schematic diagram.Studying progress arrangement view is used to see studying progress of whole students within past each week.The study into
Open up in arrangement view, the learning behavior data of each user are shown using three features: operation number accounting, accumulative study trifle account for
Than, video playback time accounting, and using the prediction result of the different color mark users.
Specific such as Fig. 7 (a) and Fig. 7 (b) is shown, and studying progress of the student in one week is shown with scatter plot.Each student
Circle on a corresponding two-dimensional surface, a circle at (x, y) represent accumulative study trifle accounting x%, operation number accounting
A student of y%.Round radius and video playback time ratio are positively correlated.What color then indicated whether to learn student's next week
Prediction result, red, which represents next week, learning behavior, and blue represents next week without learning behavior.Current cycle and this week enliven number
In the background with larger character size mark, it will be apparent that.When mouse is suspended on circle, it may appear that a suspension prompting frame is shown should
The current self-study behavior portrait of student and prediction result.
There is a time line traffic control tool on the right side of scatter plot, user dissipates by clicking the selection of the corresponding node on timeline
The cycle that point diagram is shown.In order to show that user's studying progress changes with time situation, timeline devises playing function.It is single
The broadcast button below timeline is hit, the scatter plot in left side can successively change to last current week since first week.
One is provided in the scatter plot upper right corner and includes amplification, reduction, the tool box for downloading three functions, and user click is put
After big button, a rectangular area can be selected in scatter plot, as shown in Fig. 7 (a) and Fig. 7 (b), choose the model of rear reference axis
The numberical range of the rectangular area will be set to by enclosing.In this way, rectangular area is just exaggerated, will become holding the circle in region
It must disperse, what can be more clear observing and selecting, and can also download the scatter plot of current cycle.For the weight of scatter plot
Folded problem is alleviated using two skills: first skill is the random component that a very little is superimposed in transverse and longitudinal coordinate, this makes
There must be the learner of same accumulative study trifle accounting will not be completely overlapped.Although this will affect the accuracy of round position,
But it will not influence round approximate location, the overall impression being distributed for school's studying progress will not be influenced, user is still
The exact value of learner's various features can be so chosen by underlapped region.Second skill is exactly transparent by what is justified
Degree is set as 0.5, this can make overlapping become visible.
As shown in Fig. 8 (a) and Fig. 8 (b), Fig. 8 (a) and Fig. 8 (b) is the partial schematic diagram of personal footprint view in Fig. 4.It is a
People's footprint view passes through the pop-up of triggering graphic element, individual's footprint in the studying progress arrangement view of Fig. 7 (a) and Fig. 7 (b)
The investment degree of user's study is shown in view in the form of stacking histogram, the investment degree of user study includes the use
Family is watching weekly the number of video and is submitting the number of operation.
As shown in Fig. 8 (a), when mouse-over is above some column, it will a suspension prompting frame occur and show use
Behavioral characteristics of the family in this week.In order to help user understand learner's active degree variation tendency, to viewing video number and
The sum of number of jobs is submitted to carry out cubic polynomial fitting, by matched curve and histogram Overlapping display.As shown in Fig. 8 (b), with
Studying progress arrangement view is the same, provides a tool box in the upper right corner and also increases other than amplification, reduction, download function
One display data function.By clicking " display data " button, Data View can be entered, list column in a tabular form
Scheme the data shown.
In the application scenarios, visual ways of presentation is designed to the result of data analysis, and can visual analysis.Data
Visualization is presented data in the form that user is easy to perceive, it is intended to whole numbers as far as possible completely, really, objectively be presented
According to, or even certain details therein can be enhanced.Visual analysis refers on the basis of by data visualization, by providing friendship
Mutual property tool auxiliary user independently explores and analyzes data.
Referring to Fig. 9, Fig. 9 is the structural schematic diagram of an embodiment of learning behavior dynamic prediction device of the present invention.Such as
Shown in Fig. 9, which includes neural network model generation module 90, learning behavior dynamic prediction module
91, data statistic analysis model generation module 92 and visible system generation module 93.
Neural network model generation module 90 is used for the learning behavior data based on user in historical time section, using length
When memory Recognition with Recurrent Neural Network be trained as prediction model, obtain optimal models.Learning behavior dynamic prediction module 91 is used
In for obtaining the user in the learning behavior data of first time period, the optimal models are based on, prediction result is exported.Data system
It counts analysis model generation module 92 to be used to statistically analyze based on the learning behavior data and prediction result, obtains user information number
According to.Visible system generation module 93 is used to show the user information data using visualization system.
Wherein, module 90 and module 91 become minimum working cell together, can be performed to realize above-mentioned study row
For dynamic prediction method, the working method of specific module 90 and module 91 can be found in above-mentioned, and details are not described herein.92 He of module
Module 93, which is performed, shows user data in a manner of visual analyzing to realize, working method is also together referring to above-mentioned, herein not
It repeats.
The present invention also provides a kind of learning behavior Dynamic Forecasting System equipment, which includes: at least one processor；With
And the memory being connect at least one processor communication；Wherein, be stored with can be by least one processor for the memory
The instruction of execution, which is executed by least one processor, so that at least one processor is able to carry out as above-mentioned
Prediction technique.
The present invention also provides a kind of computer readable storage medium, which has computer can
It executes instruction, the computer executable instructions are for making computer execute above-mentioned prediction technique.
In conclusion being in contrast to the prior art, the present invention is pre- as prediction model by using LSTM-RNN
Output is surveyed as a result, the accuracy of prediction can be improved, and user information data are shown using interactivity visual analysis mode, is conveniently existed
The backstage of system is checked, analyzes user data information, thus help teacher that can timely, intuitively know data analysis result, into
And realize " closed loop " of teaching process.
It is to be illustrated to preferable implementation of the invention, but the invention is not limited to the implementation above
Example, those skilled in the art can also make various equivalent variations on the premise of without prejudice to spirit of the invention or replace
It changes, these equivalent deformations or replacement are all included in the scope defined by the claims of the present application.
Claims (19)
1. a kind of learning behavior dynamic prediction method characterized by comprising
Based on user historical time section learning behavior data, using long short-term memory Recognition with Recurrent Neural Network as prediction model
It is trained, obtains optimal models；
The user is obtained in the learning behavior data of first time period, the optimal models is based on, exports prediction result.
2. learning behavior dynamic prediction method according to claim 1, which is characterized in that described to be based on user in history
Between section history learning behavioral data, be trained, obtained most as prediction model using long short-term memory Recognition with Recurrent Neural Network
The step of excellent model includes:
User is obtained in the learning behavior data of the historical time section；
Learning characteristic vector [x is extracted based on the learning behavior data1, x2..., xK] and label [y1, y2..., yK], wherein xi
For the corresponding learning characteristic vector of learning behavior data of user's i-th of unit time in the historical time section, yiFor
Indicate whether user in the i+1 unit time has learning behavior, if any learning behavior, then yi=1, conversely, then yi=0；Its
In, the K of i=1,2,3 ..., K >=1；
With learning characteristic vector [x1, x2..., xK] it is input, with [y1, y2..., yK] be label training pattern, obtain it is described most
Excellent model.
3. learning behavior dynamic prediction method according to claim 2, which is characterized in that described with learning characteristic vector
[x1, x2..., xK] it is input, [y1, y2..., yK] it is label training pattern, it is the step of obtaining the optimal models, specific to wrap
It includes:
Multiple users are divided into training set according to 4: 1 ratio using 5 folding cross-validation methods and verifying collects, and utilize grid search
Method carries out arameter optimization, obtains the optimal models.
4. learning behavior dynamic prediction method according to claim 2 or 3, which is characterized in that the long short-term memory
When Recognition with Recurrent Neural Network training, using intersection entropy function as loss function.
5. learning behavior dynamic prediction method according to claim 2 or 3, which is characterized in that the long short-term memory
When Recognition with Recurrent Neural Network training, parameter optimization is carried out using Adam method.
6. learning behavior dynamic prediction method according to claim 5, which is characterized in that the parameter includes weight vectors
And bias vector, the weight vectors initialization value are that [- 0.1,0.1] is uniformly distributed, the bias vector initialization value is zero
Vector.
7. learning behavior dynamic prediction method according to claim 2 or 3, which is characterized in that use Logistic
Activation primitive of the Sigmoid function as the long short-term memory Recognition with Recurrent Neural Network.
8. according to claim 1, learning behavior dynamic prediction method described in 2,3 or 6, which is characterized in that further include AUC verifying
Step:
Select at least one set of baseline model；
The baseline model and the optimal models are respectively adopted to verify verifying collection, obtain respective evaluation index
AUC value.
9. learning behavior dynamic prediction method according to claim 1, which is characterized in that further include:
It is statisticallyd analyze based on the learning behavior data and prediction result, obtains user information data；
The user information data are shown using visualization system.
10. learning behavior dynamic prediction method according to claim 9, which is characterized in that the visualization system uses
B/S framework is realized, using MySQL database storing data, builds website using Django frame, based on SVG element in HTML
It draws a diagram in webpage.
11. learning behavior dynamic prediction method according to claim 9 or 10, which is characterized in that the visualization system
The view shown based on user information data is included student information view, enlivens calendar view, studying progress arrangement view, individual
Footprint view.
12. learning behavior dynamic prediction method according to claim 11, which is characterized in that in the student information view
Gender, age or the distribution of receiving an education of user are described using cake chart.
13. learning behavior dynamic prediction method according to claim 11, which is characterized in that the calendar view that enlivens is
Daily any active ues number is shown on a calendar figure, each rectangle in the calendar figure represents in one day, one week
Be placed in same row within seven days, and day updated as first day of each column using course material, the depth of rectangle color with
Any active ues number direct proportionality on the same day.
14. learning behavior dynamic prediction method according to claim 11, which is characterized in that the studying progress distribution view
The learning behavior data of each user are shown using three features in figure: operation number accounting, accumulative study trifle accounting, video are broadcast
Put time accounting, and using the prediction result of the different color mark users.
15. learning behavior dynamic prediction method according to claim 11, which is characterized in that individual's footprint view is logical
It crosses and triggers graphic element pop-up in the studying progress arrangement view, using stacking histogram in individual's footprint view
Form shows that the investment degree of user's study, the investment degree of user's study include that the user is watching weekly video
Number and the number for submitting operation.
16. a kind of learning behavior dynamic prediction device characterized by comprising
Neural network model generation module, for, in the learning behavior data of historical time section, being remembered in short-term using length based on user
Recall Recognition with Recurrent Neural Network to be trained as prediction model, obtains optimal models；
Learning behavior dynamic prediction module, for obtaining the user in the learning behavior data of first time period, based on described
Optimal models export prediction result.
17. learning behavior dynamic prediction device according to claim 16, which is characterized in that further include:
Data statistic analysis model generation module is obtained for being statisticallyd analyze based on the learning behavior data and prediction result
User information data；
Visible system generation module, for showing the user information data using visualization system.
18. a kind of learning behavior Dynamic Forecasting System equipment characterized by comprising
At least one processor；And
The memory being connect at least one described processor communication；Wherein,
The memory is stored with the instruction that can be executed by least one described processor, and described instruction is by described at least one
It manages device to execute, so that at least one described processor is able to carry out such as the described in any item methods of claim 1 to 15.
19. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has computer can
It executes instruction, the computer executable instructions are for making computer execute such as the described in any item methods of claim 1 to 15.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811144725.2A CN109272164B (en) | 2018-09-29 | 2018-09-29 | Learning behavior dynamic prediction method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811144725.2A CN109272164B (en) | 2018-09-29 | 2018-09-29 | Learning behavior dynamic prediction method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109272164A true CN109272164A (en) | 2019-01-25 |
| CN109272164B CN109272164B (en) | 2021-09-28 |
Family
ID=65198830
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811144725.2A Active CN109272164B (en) | 2018-09-29 | 2018-09-29 | Learning behavior dynamic prediction method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109272164B (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109816178A (en) * | 2019-02-21 | 2019-05-28 | 四川大学 | Psychological condition prediction technique, device and electronic equipment |
| CN109858806A (en) * | 2019-01-30 | 2019-06-07 | 网易（杭州）网络有限公司 | Method, apparatus, medium and the electronic equipment of data processing |
| CN110059716A (en) * | 2019-03-12 | 2019-07-26 | 西北大学 | A kind of building of CNN-LSTM-SVM network model and MOOC discontinue one's studies prediction technique |
| CN110110939A (en) * | 2019-05-15 | 2019-08-09 | 杭州华网信息技术有限公司 | The academic record prediction and warning method of behavior is serialized based on deep learning student |
| CN110276246A (en) * | 2019-05-09 | 2019-09-24 | 威比网络科技(上海)有限公司 | Course index detects alarm method, device, electronic equipment, storage medium |
| CN110705799A (en) * | 2019-10-10 | 2020-01-17 | 北京小米移动软件有限公司 | Method, device and medium for intelligently prompting combing and washing related information |
| CN111144368A (en) * | 2019-12-31 | 2020-05-12 | 重庆和贯科技有限公司 | Student behavior detection method based on long-time and short-time memory neural network |
| CN111436939A (en) * | 2020-03-17 | 2020-07-24 | 佛山市台风网络科技有限公司 | Health monitoring method, system, computer equipment and readable storage medium |
| CN111523738A (en) * | 2020-06-22 | 2020-08-11 | 之江实验室 | System and method for predicting learning effect based on user online learning behavior pattern |
| CN111950892A (en) * | 2020-08-11 | 2020-11-17 | 华中师范大学 | Data-driven university student personalized learning behavior prediction and intervention method |
| CN112049624A (en) * | 2019-06-06 | 2020-12-08 | 中国石油天然气股份有限公司 | Method, device, equipment and storage medium for predicting dynamic reserves of oil well |
| CN112101553A (en) * | 2020-11-10 | 2020-12-18 | 鹏城实验室 | Network structure searching method and device, equipment and storage medium |
| CN112137834A (en) * | 2019-06-27 | 2020-12-29 | 丰田自动车株式会社 | Learning system, rehabilitation support system, method, program, and learning completion model |
| CN112734105A (en) * | 2021-01-08 | 2021-04-30 | 浙江工业大学 | Method for preventing breaking behavior in online education |
| CN112825147A (en) * | 2019-11-21 | 2021-05-21 | 亿度慧达教育科技（北京）有限公司 | Learning path planning method, device, equipment and storage medium |
| CN114130038A (en) * | 2021-11-25 | 2022-03-04 | 浙江小族智能科技有限公司 | Amusement vehicle and control method and device thereof, storage medium and terminal |
| CN114567815A (en) * | 2022-01-20 | 2022-05-31 | 清华大学 | Pre-training-based admiration class self-adaptive learning system construction method and device |
| CN116227729A (en) * | 2023-03-20 | 2023-06-06 | 广州工程技术职业学院 | Learning evaluation method, device and equipment based on machine learning |
| CN116362418A (en) * | 2023-05-29 | 2023-06-30 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130325560A1 (en) * | 2012-06-05 | 2013-12-05 | Rank Miner, Inc. | System, Method and Apparatus for Voice Analytics of Recorded Audio |
| CN106886846A (en) * | 2017-04-26 | 2017-06-23 | 中南大学 | A kind of bank outlets' excess reserve Forecasting Methodology that Recognition with Recurrent Neural Network is remembered based on shot and long term |
| CN107180284A (en) * | 2017-07-07 | 2017-09-19 | 北京航空航天大学 | A kind of SPOC student based on learning behavior feature shows weekly Forecasting Methodology and device |
| CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
| CN107622329A (en) * | 2017-09-22 | 2018-01-23 | 深圳市景程信息科技有限公司 | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales |
-
2018
- 2018-09-29 CN CN201811144725.2A patent/CN109272164B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20130325560A1 (en) * | 2012-06-05 | 2013-12-05 | Rank Miner, Inc. | System, Method and Apparatus for Voice Analytics of Recorded Audio |
| CN106886846A (en) * | 2017-04-26 | 2017-06-23 | 中南大学 | A kind of bank outlets' excess reserve Forecasting Methodology that Recognition with Recurrent Neural Network is remembered based on shot and long term |
| CN107180284A (en) * | 2017-07-07 | 2017-09-19 | 北京航空航天大学 | A kind of SPOC student based on learning behavior feature shows weekly Forecasting Methodology and device |
| CN107423442A (en) * | 2017-08-07 | 2017-12-01 | 火烈鸟网络(广州)股份有限公司 | Method and system, storage medium and computer equipment are recommended in application based on user's portrait behavioural analysis |
| CN107622329A (en) * | 2017-09-22 | 2018-01-23 | 深圳市景程信息科技有限公司 | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109858806A (en) * | 2019-01-30 | 2019-06-07 | 网易（杭州）网络有限公司 | Method, apparatus, medium and the electronic equipment of data processing |
| CN109816178A (en) * | 2019-02-21 | 2019-05-28 | 四川大学 | Psychological condition prediction technique, device and electronic equipment |
| CN110059716A (en) * | 2019-03-12 | 2019-07-26 | 西北大学 | A kind of building of CNN-LSTM-SVM network model and MOOC discontinue one's studies prediction technique |
| CN110276246A (en) * | 2019-05-09 | 2019-09-24 | 威比网络科技(上海)有限公司 | Course index detects alarm method, device, electronic equipment, storage medium |
| CN110110939A (en) * | 2019-05-15 | 2019-08-09 | 杭州华网信息技术有限公司 | The academic record prediction and warning method of behavior is serialized based on deep learning student |
| CN112049624B (en) * | 2019-06-06 | 2024-04-30 | 中国石油天然气股份有限公司 | Method, device, equipment and storage medium for predicting dynamic reserve of oil well |
| CN112049624A (en) * | 2019-06-06 | 2020-12-08 | 中国石油天然气股份有限公司 | Method, device, equipment and storage medium for predicting dynamic reserves of oil well |
| CN112137834A (en) * | 2019-06-27 | 2020-12-29 | 丰田自动车株式会社 | Learning system, rehabilitation support system, method, program, and learning completion model |
| CN110705799A (en) * | 2019-10-10 | 2020-01-17 | 北京小米移动软件有限公司 | Method, device and medium for intelligently prompting combing and washing related information |
| CN112825147B (en) * | 2019-11-21 | 2022-08-26 | 亿度慧达教育科技（北京）有限公司 | Learning path planning method, device, equipment and storage medium |
| CN112825147A (en) * | 2019-11-21 | 2021-05-21 | 亿度慧达教育科技（北京）有限公司 | Learning path planning method, device, equipment and storage medium |
| CN111144368A (en) * | 2019-12-31 | 2020-05-12 | 重庆和贯科技有限公司 | Student behavior detection method based on long-time and short-time memory neural network |
| CN111436939A (en) * | 2020-03-17 | 2020-07-24 | 佛山市台风网络科技有限公司 | Health monitoring method, system, computer equipment and readable storage medium |
| CN111523738A (en) * | 2020-06-22 | 2020-08-11 | 之江实验室 | System and method for predicting learning effect based on user online learning behavior pattern |
| CN111950892B (en) * | 2020-08-11 | 2023-08-18 | 华中师范大学 | Data-driven college student personalized learning behavior prediction and intervention method |
| CN111950892A (en) * | 2020-08-11 | 2020-11-17 | 华中师范大学 | Data-driven university student personalized learning behavior prediction and intervention method |
| CN112101553A (en) * | 2020-11-10 | 2020-12-18 | 鹏城实验室 | Network structure searching method and device, equipment and storage medium |
| CN112734105A (en) * | 2021-01-08 | 2021-04-30 | 浙江工业大学 | Method for preventing breaking behavior in online education |
| CN114130038A (en) * | 2021-11-25 | 2022-03-04 | 浙江小族智能科技有限公司 | Amusement vehicle and control method and device thereof, storage medium and terminal |
| CN114130038B (en) * | 2021-11-25 | 2023-08-22 | 浙江小族智能科技有限公司 | Amusement vehicle, control method and device thereof, storage medium and terminal |
| CN114567815B (en) * | 2022-01-20 | 2023-05-02 | 清华大学 | Pre-training-based adaptive learning system construction method and device for lessons |
| CN114567815A (en) * | 2022-01-20 | 2022-05-31 | 清华大学 | Pre-training-based admiration class self-adaptive learning system construction method and device |
| CN116227729A (en) * | 2023-03-20 | 2023-06-06 | 广州工程技术职业学院 | Learning evaluation method, device and equipment based on machine learning |
| CN116227729B (en) * | 2023-03-20 | 2023-10-10 | 广州工程技术职业学院 | Learning evaluation method, device and equipment based on machine learning |
| CN116362418A (en) * | 2023-05-29 | 2023-06-30 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
| CN116362418B (en) * | 2023-05-29 | 2023-08-22 | 天能电池集团股份有限公司 | Online prediction method for application-level manufacturing capacity of intelligent factory of high-end battery |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109272164B (en) | 2021-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109272164A (en) | Learning behavior dynamic prediction method, device, equipment and storage medium | |
| Schlüter et al. | A framework for mapping and comparing behavioural theories in models of social-ecological systems | |
| An | Modeling human decisions in coupled human and natural systems: Review of agent-based models | |
| Glass | Primary, secondary, and meta-analysis of research | |
| CN108596420A (en) | A kind of talent assessment system and method for Behavior-based control | |
| Park et al. | A multidimensional IRT approach for dynamically monitoring ability growth in computerized practice environments | |
| Mao et al. | Analysis of entrepreneurship education in colleges and based on improved decision tree algorithm and fuzzy mathematics | |
| Jinfeng et al. | Design of evaluation system of physical education based on machine learning algorithm and SVM | |
| Antelmi et al. | Analyzing, exploring, and visualizing complex networks via hypergraphs using SimpleHypergraphs. jl | |
| Farahbakhsh et al. | Modelling coupled human–environment complexity for the future of the biosphere: strengths, gaps and promising directions | |
| Wang et al. | A data mining approach for training evaluation in simulation-based training | |
| Efrati et al. | A data mining approach to the analysis of students’ learning styles in an e-learning community: A case study | |
| Varade et al. | Academic performance prediction of undergraduate students using decision tree algorithm | |
| Wang | [Retracted] Research on the Application of Genetic Algorithm in Physical Education | |
| Oztekin et al. | A business-analytic approach to identify critical factors in quantitative disciplines | |
| Peng | An effective analysis of online education model based on artificial intelligence | |
| Jiang et al. | ATSA: An Adaptive Tree Seed Algorithm based on double-layer framework with tree migration and seed intelligent generation | |
| Rahimi et al. | Learning analytics dashboards in educational games | |
| Wang et al. | Analysis and Prediction of Satisfaction Index of Online Learning | |
| Peng et al. | Construction and Application of Farmers’ Practical Teaching System in Vocational Education Based on Big Data Mining Technology | |
| Fernández et al. | Exploring approaches to educational data mining and learning analytics, to measure the level of acquisition of student's learning outcome | |
| Pangarkar et al. | Artificial intelligence based analysis of student’s involvement in edtech industry | |
| Yu et al. | Unsupervised Representations Predict Popularity of Peer-Shared Artifacts in an Online Learning Environment | |
| Liu | Research on the Impact of Elective Subjects of College Entrance Examination on Talent Cultivation of Higher Vocational Colleges Based on Students’ Cognitive Behavioral Characteristics | |
| Fu et al. | Design and research of educational mode in context of teaching gamification |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |