CN106157204A - A kind of Forecasting The Scores method and system based on BP neural network model - Google Patents
A kind of Forecasting The Scores method and system based on BP neural network model Download PDFInfo
- Publication number
- CN106157204A CN106157204A CN201610443524.7A CN201610443524A CN106157204A CN 106157204 A CN106157204 A CN 106157204A CN 201610443524 A CN201610443524 A CN 201610443524A CN 106157204 A CN106157204 A CN 106157204A
- Authority
- CN
- China
- Prior art keywords
- student
- data
- achievement
- learning
- school
- 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 54
- 238000003062 neural network model Methods 0.000 title claims abstract description 20
- 230000007935 neutral effect Effects 0.000 claims abstract description 53
- 238000012549 training Methods 0.000 claims abstract description 35
- 238000006243 chemical reaction Methods 0.000 claims abstract description 21
- 230000008569 process Effects 0.000 claims description 31
- 230000006399 behavior Effects 0.000 claims description 9
- 238000013480 data collection Methods 0.000 claims description 8
- 210000001747 pupil Anatomy 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 210000002569 neuron Anatomy 0.000 description 9
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000000694 effects Effects 0.000 description 5
- 238000005516 engineering process Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000012897 Levenberg–Marquardt algorithm Methods 0.000 description 1
- 238000003556 assay Methods 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 210000004218 nerve net Anatomy 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
- 239000013598 vector 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
- 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
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Human Resources & Organizations (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Primary Health Care (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of Forecasting The Scores method and system based on BP neural network model, by obtaining student learning data, and the data obtained are transmitted to the data base of first server；For the student learning data of storage in data base, carry out data conversion, obtain normalized Students ' Learning status data table；Each property value for normalized Students ' Learning status data table carries out orthogonal coding, builds the training dataset of BP neutral net, carries out structure and the training of BP neutral net result prediction model based on this training dataset；Input after student data to be predicted is carried out data conversion, standardization, orthogonal coding and normalized to the result prediction model based on BP neutral net trained carry out school achievement classification prediction, it is thus achieved that Forecasting The Scores result is also shown by display unit.The BP neural network model that the present invention uses, described model is for realizing the accurate prediction of Scores, and reliability is high.
Description
Technical field
The present invention relates to education skill, computer utility ambit, be specifically related to a kind of based on BP neural network model
Forecasting The Scores method and system.
Background technology
At present, China Higher educational development is rapid, and wherein the quantity of regular higher educational institutions has reached more than 2500.In recent years
Come, the continuous decrease of source of students, bring huge crisis of survival to regular higher educational institutions.How to improve students developing quality, carry
The employment competition ability of high student, becomes many colleges and universities and needs a problem of solution badly.The school achievement of student is trained as student
Support an important core index of quality, the extremely concern of university managementt person.
China Higher universities and colleges are generally responsible for the daily management of student by Youth League committee counsellor (form master), are responsible for student by teacher
Course is theoretical and technical ability training.Often lacking effective communication exchange between counsellor and teacher, this can cause one
Divide classmate to go astray because being negligent of supervisor, be forced because school achievement is the best to delay to graduate or leave school.The school achievement of student
Suffer from many impacts, the factors such as including the conventional achievement of student, learning capacity, teacher's guidance situation.If energy
Reaching the history school achievement according to student and the general performance of each side and quality situation, the following school achievement to student is carried out
Prediction；And predicting the outcome according to school achievement, education that the student being likely to occur problem is strengthened management in time, supervise it conscientious
Study, with avoid its occur cannot be by the consequence of academic examination, this will significantly facilitate counsellor's educational management for student,
To play an important role for improving the training quality of student.
Although current all kinds of Education Administration Information System is the most universal in institution of higher learning, it is possible to enter student learning achievement
The effective management of row.But, the research work being predicted the school achievement of student analyzing is the most rare, also has not seen wide
General enforcement.
1, existing Education Administration Information System, only lays particular emphasis on the management for Students ' Learning achievement data, and ignores
The management of other behavioral data raw.To the collection of student data imperfect, it is also difficult to student is carried out comprehensive assay.
2, for student achievement data, it is only that student performance is entered into Education Administration Information System at present；Teaching management system
What system stored is all the historical data of student performance.It is evaluated being merely by student performance to the existing ability of student
Historical data analysis obtain, be provided without corresponding data processing model, it is impossible to realize intelligent predicting to Scores.
The technical problem that patent of the present invention exists when the prediction of Scores for these just, by data mining
Technology is applied to the prediction of school achievement, it is achieved a kind of Forecasting The Scores method based on BP neural network model and be
System, makes every effort to promote the development of this research.
Summary of the invention
For solving the deficiency that prior art exists, the invention discloses a kind of student's school work based on BP neural network model
Result prediction method and system, the present invention realizes Scores by using corresponding data acquisition and analytical technology
Intelligent predicting.
For achieving the above object, the concrete scheme of the present invention is as follows:
A kind of Forecasting The Scores method based on BP neural network model, comprises the following steps:
Step one: obtain student learning data, and the data obtained are transmitted to the data base of first server；
Step 2: for the student learning data of storage in data base, carry out data conversion, obtain normalized student
Study condition tables of data；
Step 3: each property value of normalized Students ' Learning status data table is carried out orthogonal coding, builds BP neural
The training dataset of network, carries out structure and the training of the result prediction model of BP neutral net based on this training dataset；
Step 4: defeated after student data to be predicted is carried out data conversion, standardization, orthogonal coding and normalized
Enter to the result prediction model based on BP neutral net trained carry out school achievement classification prediction, it is thus achieved that Scores
Predict the outcome and shown by display unit.
Further, in described step one, described student learning data include the school achievement information of student, its middle school
Industry achievement is the information of storage in Education Administration Information System database server, Education Administration Information System database server and the first clothes
Business device communicates, and the school achievement of student is transmitted to first server.
Further, described school achievement information includes that student's school achievement adjacent to two terms becomes with admission school work
Achievement, wherein the school achievement situation of front halves and admission school achievement situation, using the history school achievement as individual students
Attribute；The school achievement situation of rear halves is using the classification results as Scores.
Further, in described step one, described student learning data also include learning behavior information, and learning behavior is believed
The acquisition of breath is by data collection station, and data collection station can be computer or removable smart machine.
Further, described learning behavior information specifically includes learning time, net fun time, library use frequency
And borrow books type etc..
Further, when specifically obtaining of Scores, utilize the student number of student for term from teaching management system
System database server extracts class's list of results of this student achievement data and its place class.
Further, in first server, student data is changed, according to the interval at student information data place, will
To continuous data segmentation be converted to level data.
Further, for school achievement information, become with admission school work adjacent to the school achievement in two terms including student
Achievement situation, these data need to carry out conversion process, concrete handling process:
Obtain class's list of results, according to the test subject quantity of student, calculate the average achievement of student, and by student's
Average achievement sorts, output class ranking list；And export class's pupil load；
According to class's ranking list and student achievement data, the ranking of inquiry student, and export.
According to student's ranking and class's pupil load, it is judged that student's ranking overall positions in class.
If belonging to front 20%, then output Scores grade is A；If located in after 20%, before 40%, then
Output Scores grade is B；If located in after 40%, before 60%, then output Scores grade is C；
If located in after 60%, before 80%, then output Scores grade is D；If located in rear 20%, then export student
School achievement grade is E.
Further, in step 3, each property value in the normalized Students ' Learning status data table obtained is entered
The detailed process of row orthogonal coding includes:
Travel through normalized Students ' Learning status data table, therefrom add up the number of the property value of a certain attribute, export N；
N number of binary digit is used to carry out orthogonal coding, and output orthogonal coding the property value of current attribute.
Further, in step 3, the detailed process of the training dataset building BP neutral net includes:
Student's record is taken one by one from normalized Students ' Learning status data table；
The orthogonal coding of the property value of each attribute of student's record is merged, constitutes student's note of an orthogonal coding
Record；
It is normalized, corresponding to remaining field orthogonal coding sets the orthogonal coding position corresponding to field
Orthogonal coding invariant position, obtains student's record of normalized, builds the training dataset of normalized.
Further, in step 3, build result prediction model based on BP neutral net, determine that BP neutral net has
Having three layers, i.e. input layer, hidden layer and output layer, the method to set up of each parameter value of BP neutral net is as follows:
(1) determine BP neural network input layer neuron number, set input layer number as normalized student
Study condition tables of data sets the length of the orthogonal coding value of field correspondence attribute；
(2) determine BP neutral net output layer neuron number, set output layer neuron number as normalized student
The length of the orthogonal coding value of remaining field attribute in study condition tables of data；
(3) BP neutral net hidden neuron number is determined；
(4) activation primitive of each layer of BP neutral net is determined；
(5) BP neural network training method is determined；
(6) the training requirement precision of BP neutral net is determined；
(7) learning rate of BP neutral net is determined.
A kind of Forecasting The Scores system based on BP neural network model, including:
Data acquisition module: obtain student learning data, and the data obtained are transmitted the data to first server
In storehouse；
Data conversion module: for the student learning data of storage in data base, carry out data conversion, standardized
Students ' Learning status data table；
Result prediction model building module based on BP neutral net: for normalized Students ' Learning status data table
Each property value carries out orthogonal coding, builds the training dataset of BP neutral net, carries out BP nerve net based on this training dataset
The structure of the result prediction model of network and training；
Student's school work prediction module: student data to be predicted is carried out data conversion, standardization, orthogonal coding and normalizing
Change process after input to the result prediction model based on BP neutral net trained carry out school achievement classification prediction, it is thus achieved that learn
Raw school achievement is predicted the outcome and is shown by display unit.
Beneficial effects of the present invention:
1, the present invention proposes the Forecasting The Scores method and system of result prediction model based on BP neutral net,
Following school achievement of measurable student, in order to institution of higher learning strengthen the educational management to student.
2, the present invention describes student's sample by 14 attributes such as the history school achievement of student and learning behavior information,
Its sample data can be obtained by the Education Administration Information System of school and data collection station, and its Data Source is easy and accurate, it is simple to
It is widely popularized in institution of higher learning.
3, the data that the present invention is directed to obtain carry out data conversion, obtain the Students ' Learning status data table of specification, according to
The interval at the place of student information data, is converted to level data by its segmentation, to reduce the quantity of property value, it is simple to follow-up mould
The utilization of data when type is set up.
4, the data message of the student of acquisition is all stored to the data base of first server by the present invention, in order to follow-up number
According to call and process conveniently, and ensure safety and the stability of data.
5, the student-directed task of current institution of higher learning counsellor is heavy, it is difficult to respect to each student, the present invention is by energy
Enough effectively doping, for counsellor, the student that school work will go wrong, this, for strengthening the specific aim of counsellor work, carries
The students developing quality of high institution of higher learning will play useful effect.
6. the result prediction model based on BP neutral net that the present invention uses, described model is for realizing student performance
Accurately prediction, reliability is high.
Accompanying drawing explanation
The overall prediction flow chart of Fig. 1 present invention；
The Scores of Fig. 2 present invention specifically obtain flow chart；
The school achievement information data of Fig. 3 present invention carries out the concrete process chart of conversion process；
Each property value in Fig. 4 present invention normalized Students ' Learning status data table, carries out the concrete stream of orthogonal coding
Cheng Tu；
The particular flow sheet of the training dataset of the structure BP neutral net of Fig. 5 present invention.
Detailed description of the invention:
The present invention is described in detail below in conjunction with the accompanying drawings:
As it is shown in figure 1, the Forecasting The Scores method based on BP neural network model of the present invention includes following master
Want step:
Step one: be collected student data, including school achievement and the learning behavior information of student；
Step 2: student data is carried out conversion process, obtains the Students ' Learning status data table of specification；
Step 3: by the Students ' Learning status data table of specification, trains Scores based on BP neutral net pre-
Survey model；
Step 4: the result prediction model based on BP neutral net obtained according to step 3, it was predicted that the school work of student
Achievement classification.
Utilize the present invention, can be according to the daily study condition of student, it was predicted that following school achievement of student, it is possible to just
The educational management to student is strengthened in institution of higher learning.
Wherein, in step one, the school achievement of student directly can be derived by Education Administration Information System data base, and other learns row
For information etc. or obtain (also can provide e-survey questionnaire by means of network) by student is carried out questionnaire or adopt
Obtaining with data collection station, preferably select the mode of data collection station to obtain, data collection station is terminal
Or movable equipment, using student number as the ID of student, each student all has and only one of which student number, there is phase under each student number
The data record answered, because the data so obtained can the just fraud that may be present of face questionnaire and the distortion that causes
The problem of data.
When by network provide e-survey questionnaire obtain data time, the student number of e-survey questionnaire and student one a pair
Should, the content each student filled in carries out unifying to collect and process.
For other learning behavior information may include that the Students ' Learning time, on time attend class situation, the net fun time,
Library uses frequency, borrows books type, Time Management, learning capacity, extracurricular activities, teacher's guidance situation, family
Instruct situation, special interest.
Such as, when the Students ' Learning time starts, computer starts timing, at the end of the Students ' Learning time, and computer timing
Terminate, then obtain this student learning time；
Attend class situation on time, by the way of fingerprint recognition, each student at school time carry out fingerprint recognition, do not carry out
The student of fingerprint recognition is then absent from school, obtains the situation of on time attending class of student in this way.
The acquisition of net fun time is similar with the acquisition mode of Students ' Learning time.
Library uses frequency, borrowing books type can be by the database server in the book management system of school
Middle acquisition, by the data transmission of storage in the database server in book management system to computer.
The number such as Time Management, learning capacity, extracurricular activities, teacher's guidance situation, familial Guide situation, special interest
According to can comprehensively individual self evaluation and teacher, the others'evaluation of classmate and determine.
The attribute of all kinds of student informations is as shown in table 1.
Table 1
For school achievement information, collect student adjacent to the school achievement in two terms and admission school achievement situation, its
The school achievement situation of halves and admission school achievement situation before in, using the history school achievement attribute as individual students；
The school achievement situation of rear halves is using the classification results as Scores.The concrete acquisition of various Scores
Step is the most as shown in Figure 2.Student number according to student and the student achievement data storehouse of school, therefrom extract student achievement data and
Class's list of results of its place class.
In step 2, for school achievement information, become with admission school work adjacent to the school achievement in two terms including student
Achievement situation, these data need to carry out conversion process.Concrete handling process is as shown in Figure 3.
To the class's list of results obtained, according to the test subject quantity of student, calculate the average achievement of student, and by learning
Raw average achievement sequence, output class ranking list；And export class's pupil load.
According to class's ranking list and student achievement data, the ranking of inquiry student, and export.
According to student's ranking and class's pupil load, it is judged that student's ranking overall positions in class.If before belonging to
20%, then output Scores grade is A；If located in after 20%, before 40%, then output Scores etc.
Level is B；If located in after 40%, before 60%, then output Scores grade is C；If located in after 60%,
Before 80%, then output Scores grade is D；If located in rear 20%, then output Scores grade is E.
School achievement information after other learning behavior information, conversion is combined, obtains Students ' Learning situation number
According to table.
In this embodiment it is assumed that obtain Students ' Learning status data table as shown in table 2.
Table 2
In step 3, by the Students ' Learning status data table of specification, train Scores based on BP neutral net
Forecast model.
Field " the school achievement situation of rear halves " in table 2 is denoted as R by the present invention14；By 13 words of other in table 2
Section, is denoted as R successively1～R13。
Step 3.1: to each property value in the study condition tables of data of the specification by step 2 gained, carry out orthogonal volume
Code.
Step 3.1 be embodied as flow process as shown in Figure 4.
Traversal Students ' Learning status data table, therefrom adds up the number of the property value of a certain attribute, exports N.
N number of binary digit is used to carry out orthogonal coding, and output orthogonal coding the property value of current attribute.
In the present embodiment, for " learning time " attribute, when adding up the number of property value of a certain attribute, can be obtained it
Property value is respectively ">5 ", " 5 ", " 4 ", " 3 ", "<3 ", totally 5 property values, therefore N is 5.When using N number of binary digit to each genus
When property value encodes, use 5 binary digits that 5 property values are encoded respectively.Property value " > 5 ", " 5 ", " 4 ", " 3 ",
" < 3 " by be encoded as respectively " 10000 ", " 01000 ", " 00100 ", " 00010 ", " 0000
1 ", these 5 binary coding vectors are obviously all orthogonal.
In like manner, for " situation of attending class " attribute on time, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively
" A ", " B ", " C ", " D ", " E ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 001
0 0”、“0 0 0 1 0”、“0 0 0 0 1”。
For " net fun time " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value be respectively " < 1 ",
" 1 ", " 2 ", " 3 ", " > 3 ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 0010
0”、“0 0 0 1 0”、“0 0 0 0 1”。
For " library use frequency " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively " > 4 ",
" 4 ", " 3 ", " 2 ", " < 2 ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 0010
0”、“0 0 0 1 0”、“0 0 0 0 1”。
For " borrowing books type " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively " specialty book
Nationality ", " magazine ", " novel ", " other ", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 001
0”、“0 0 0 1”。
For " Time Management " attribute, its property value respectively " fabulous ", " preferably ", " typically ", " poor ", totally 4
Property value；" 1000 ", " 0100 ", " 0010 ", " 0001 " will be encoded as respectively.
For " learning capacity " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively " fabulous ", " relatively
Well ", " typically ", " poor ", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 0010 ", " 00
0 1”。
For " extracurricular activities " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively " fabulous ", " relatively
Well ", " typically ", " poor ", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 0010 ", " 00
0 1”。
For " teacher's guidance situation " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value respectively " fabulous ",
" preferably ", " typically ", " poor ", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 0010 ", " 0
0 0 1”。
For " familial Guide situation " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value respectively " fabulous ",
" preferably ", " typically ", " poor ", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 0010 ", " 0
0 0 1”。
For " special interest " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value respectively " good ", " having ",
" typically ", "None", totally 4 property values；By be encoded as respectively " 1000 ", " 0100 ", " 0010 ", " 000
1”。
For " the school achievement situation of front halves " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively
For " A ", " B ", " C ", " D ", " E ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 00
1 0 0”、“0 0 0 1 0”、“0 0 0 0 1”。
For " admission school achievement situation " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value respectively " A ",
" B ", " C ", " D ", " E ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 0010
0”、“0 0 0 1 0”、“0 0 0 0 1”。
For " rear halves school achievement situation " attribute, can obtain according to flow process described in above-mentioned Fig. 4, its property value is respectively
" A ", " B ", " C ", " D ", " E ", totally 5 property values；By be encoded as respectively " 10000 ", " 01000 ", " 001
0 0”、“0 0 0 1 0”、“0 0 0 0 1”。
Step 3.2: build the training dataset of BP neutral net.
Step 3.2 be embodied as flow process as shown in Figure 5.
It is responsible for from Students ' Learning status data storehouse, taking student's record one by one.This student records after treatment, will be as one
Bar training sample.The quantity of the sample taken freely can set according to practical situation.
The orthogonal coding of the property value of each attribute of student's record is merged, constitutes student's note of an orthogonal coding
Record.
To R in orthogonal coding1～R13Corresponding orthogonal coding position is normalized, R14Corresponding is constant,
To student's record of normalized, build the training dataset of normalized.Training data concentrates front 58 dimensions of each record
Using the 58 dimension inputs as neutral net, rear 5 dimensions are using the 5 dimension target outputs as neutral net.
In the present embodiment, as a example by the Article 1 record in table 2, the concrete grammar of step 3.2 is described.
Take out table 2 Article 1 record, it may be assumed that 4, A, 1, > 4, professional book, fabulous, fabulous, general, fabulous, general, good,
A、B、A。
The orthogonal coding of each property value of gained is encoded, such as attribute according to using N number of binary digit that each property value is carried out
Value 4 corresponding " 00010 ", property value A corresponding " 10000 " etc., can obtain the orthogonal coding corresponding to current record is: 0
0 1 0 0 1 0 0 0 0 0 1 0 0 0 1 0 0 0 0 1 0 0 0 1 0 0 0 1 0 0 0 0 0 1 0 1 0 0 0
00101000100000100010000, totally 63 binary digits, wherein attribute R1～R13Right
Answer front 58 binary digits, attribute R14Corresponding 59th～63 binary digit.
To R in orthogonal coding1～R13Corresponding orthogonal coding position (the most front 58 binary digits) is normalized,
R14Corresponding is constant, obtains the student data collection of normalized.The maximum of the most front 58 binary digits is 1,
Little value is 0, and after normalized, maximum 1 is still 1, and minima 0 translates into-1.Can obtain at current record normalization
Form after reason is :-1-1 1-1-1 1-1-1-1-1-1 1-1-1-1 1-1-1-1-1 1-1-1
-1 1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 -1 1 -1 -1 -1 -1 -1 1 -1 1 -1 -1 -1 1 -1 -1
-1 -1 -1 1 -1 -1 -1 1 0 0 0 0.Wherein, front 58 dimension will as neutral net 58 dimension input, rear 5 dimension using as
5 dimension target outputs of neutral net.
Step 3.3: build result prediction model based on BP neutral net.
Determine that BP neutral net has three layers, i.e. input layer, hidden layer and output layer.Setting of each parameter value of BP neutral net
Put method as follows:
(1) BP neural network input layer neuron number is determined.Set input layer number as R1～R13Attribute
The length of orthogonal coding value, i.e. 58.
(2) BP neutral net output layer neuron number is determined.Set output layer neuron number as R14Attribute orthogonal
The length of encoded radio, i.e. 5.
(3) BP neutral net hidden neuron number is determined.The most basic principle determining the number of hidden nodes is: in satisfied essence
Degree exhausts possible compact structure on the premise of requiring, i.e. exhaust the number of hidden nodes that may lack.At this rule of thumb, by hidden layer
Neuron number is set to 3.
(4) activation primitive of each layer of BP neutral net is determined.Logarithm S-shaped is used to turn respectively at input layer, hidden layer, output layer
Move function logsig, logarithm S-shaped transfer function logsig, linear function purelin.
(5) BP neural network training method is determined.Use Levenberg-Marquardt algorithm.
(6) the training requirement precision of BP neutral net is determined.Training requirement precision setting is 0.01.
(7) learning rate of BP neutral net is determined.Learning rate is set to 0.01.
In the present embodiment, on the training dataset that step 3.2 is obtained, use the neural network function of Matlab
Newff completes structure and the training of result prediction model based on BP neutral net.In the training process, except described above
7 parameters outside, other parameter use Matlab neutral net default parameter.Through experiment, for the little rule in the present embodiment
Mould training dataset, BP neutral net can obtain the accuracy of 100%.
Step 4, the result prediction model based on BP neutral net obtained according to step 3, it was predicted that the school work of student
Achievement classification.
It is embodied as step as described below.
Step 4.1: student's sample to be predicted is carried out data conversion, standardization, orthogonal coding and normalized.
In the present embodiment, by step 4.1, it is assumed that after the data of student sample X being carried out conversion process according to step 2,
Its learning time, on time attend class situation, the net fun time, library use frequency, borrow books type, time management energy
Power, learning capacity, extracurricular activities, teacher's guidance situation, familial Guide situation, special interest, the school achievement feelings of front halves
Condition, admission school achievement situation are respectively as follows: 3, C, 2, < 2, novel, difference, general, poor, poor, general, general, E, D.Its orthogonal volume
Code is: 000100010000100000010010000100100001
0001001000100000100010, normalized can obtain :-1-1-1 1-1-1-
1 1 -1 -1 -1 -1 1 -1 -1 -1 -1 -1 -1 1 -1 -1 1 -1 -1 -1 -1 1 -1 -1 1 -1 -1 -1
-1 1 -1 -1 -1 1 -1 -1 1 -1 -1 -1 1 -1 -1 -1 -1 -1 1 -1 -1 -1 1 -1。
Step 4.2: utilize the result prediction model based on BP neutral net that step 3 is obtained, it was predicted that student's sample
School achievement classification.
In the present embodiment, by step 4.2, the result prediction mould based on BP neutral net constructed by step 3.3 is utilized
The school achievement classification of the rear halves of student X is predicted by type, and the output valve that can obtain neutral net is [0.0508-
0.0572 0.3938 0.1865 0.4261] ', wherein " ' " representing matrix transposition.Take classification corresponding to maximum as prediction
As a result, the Scores of the output correspondence that can obtain neutral net is " E " class.
According to this student's sample school work prediction achievement, its school achievement belong to " E " class, i.e. ranking ranking will 80% it
After, it is clear that tackle this classmate and strengthen education management.Counsellor can predict the outcome according to this, is intervened this student in time, right
It is criticized and educate, and corrects its bad study habit, rectifies its learning attitude, to avoid the school achievement of this life to occur seriously asking
Topic.
It should be noted that the Forecasting The Scores of the result prediction model based on BP neutral net in the application
Method and system are carried out on the basis of being all based on the hardware products such as existing computer, server, obtained prediction knot
Fruit can be shown by corresponding display unit.
Although the detailed description of the invention of the present invention is described by the above-mentioned accompanying drawing that combines, but not the present invention is protected model
The restriction enclosed, one of ordinary skill in the art should be understood that on the basis of technical scheme, and those skilled in the art are not
Need to pay various amendments or deformation that creative work can make still within protection scope of the present invention.
Claims (10)
1. a Forecasting The Scores method based on BP neural network model, is characterized in that, comprise the following steps:
Step one: obtain student learning data, and the data obtained are transmitted to the data base of first server；
Step 2: for the student learning data of storage in data base, carry out data conversion, obtain normalized Students ' Learning
Status data table；
Step 3: each property value of normalized Students ' Learning status data table is carried out orthogonal coding, builds BP neutral net
Training dataset, carry out structure and the training of BP neutral net result prediction model based on this training dataset；
Step 4: input after student data to be predicted is carried out data conversion, standardization, orthogonal coding and normalized to
The result prediction model based on BP neutral net trained carries out school achievement classification prediction, it is thus achieved that Forecasting The Scores
Result is also shown by display unit.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
In described step one, described student learning data include the school achievement information of student, and wherein school achievement is teaching management
The information of storage in system database server, Education Administration Information System database server communicates with first server, will
The school achievement of student is transmitted to first server.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
Described school achievement information include student adjacent to the school achievement in two terms and admission school achievement, wherein before of halves
Industry achievement situation and admission school achievement situation, using the history school achievement attribute as individual students；The school work of rear halves
Achievement situation is using the classification results as Scores.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
In described step one, described student learning data also include that learning behavior information, the acquisition of learning behavior information pass through data
Acquisition terminal, data collection station can be computer or removable smart machine.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
When specifically obtaining of Scores, utilizes the student number of student to carry from Education Administration Information System database server for term
Take out class's list of results of this student achievement data and its place class.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
Student data is changed by first server, according to the interval at student information data place, the continuous data obtained is divided
Section is converted to level data.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
For school achievement information, need with admission school achievement situation, these data adjacent to the school achievement in two terms including student
Carry out conversion process, concrete handling process:
Obtain class's list of results, according to the test subject quantity of student, calculate the average achievement of student, and press the average of student
Achievement sorts, output class ranking list；And export class's pupil load；
According to class's ranking list and student achievement data, the ranking of inquiry student, and export.
According to student's ranking and class's pupil load, it is judged that student's ranking overall positions in class.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
In step 3, each property value in the normalized Students ' Learning status data table obtained is carried out the concrete mistake of orthogonal coding
Journey includes:
The Students ' Learning status data table of traversal specification, therefrom adds up the number of the property value of a certain attribute, exports N；
N number of binary digit is used to carry out orthogonal coding, and output orthogonal coding the property value of current attribute.
A kind of Forecasting The Scores method based on BP neural network model, is characterized in that,
In step 3, the detailed process of the training dataset building BP neutral net includes:
Student's record is taken one by one from normalized Students ' Learning status data table；
The orthogonal coding of the property value of each attribute of student's record is merged, constitutes student's record of an orthogonal coding；
It is normalized orthogonal coding sets the orthogonal coding position corresponding to field, orthogonal corresponding to remaining field
Bits of coded is constant, obtains student's record of normalized, builds the training dataset of normalized.
10. a Forecasting The Scores system based on BP neural network model, is characterized in that, including:
Data acquisition module: obtain student learning data, and the data obtained are transmitted to the data base of first server；
Data conversion module: for the student learning data of storage in data base, carry out data conversion, obtain normalized
Raw study condition tables of data；
Result prediction model building module based on BP neutral net: for each genus of normalized Students ' Learning status data table
Property value carry out orthogonal coding, build the training dataset of BP neutral net, carry out BP neutral net one-tenth based on this training dataset
The structure of achievement forecast model and training；
Student's school work prediction module: student data to be predicted is carried out at data conversion, standardization, orthogonal coding and normalization
Input after reason to the result prediction model based on BP neutral net trained carry out school achievement classification prediction, it is thus achieved that Xue Shengxue
Industry result prediction result is also shown by display unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610443524.7A CN106157204A (en) | 2016-06-20 | 2016-06-20 | A kind of Forecasting The Scores method and system based on BP neural network model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610443524.7A CN106157204A (en) | 2016-06-20 | 2016-06-20 | A kind of Forecasting The Scores method and system based on BP neural network model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106157204A true CN106157204A (en) | 2016-11-23 |
Family
ID=57353283
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610443524.7A Pending CN106157204A (en) | 2016-06-20 | 2016-06-20 | A kind of Forecasting The Scores method and system based on BP neural network model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106157204A (en) |
Cited By (22)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682768A (en) * | 2016-12-08 | 2017-05-17 | 北京粉笔蓝天科技有限公司 | Prediction method, system, terminal and server of test score |
| CN107726632A (en) * | 2017-09-29 | 2018-02-23 | 深圳和而泰智能控制股份有限公司 | Method for heating and controlling and server |
| CN107895033A (en) * | 2017-11-24 | 2018-04-10 | 淮阴工学院 | A kind of method for early warning of student's academic warning system based on machine learning |
| CN108171358A (en) * | 2017-11-27 | 2018-06-15 | 科大讯飞股份有限公司 | Score prediction method and device, storage medium and electronic device |
| CN108280531A (en) * | 2017-07-28 | 2018-07-13 | 淮阴工学院 | A kind of student class marks sequencing prediction technique returned based on Lasso |
| CN108320045A (en) * | 2017-12-20 | 2018-07-24 | 卓智网络科技有限公司 | Student performance prediction technique and device |
| CN109214579A (en) * | 2018-09-20 | 2019-01-15 | 山东省林业科学研究院 | Salt-soda soil stability prediction method and system based on BP neural network |
| 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 |
| CN110223553A (en) * | 2019-05-20 | 2019-09-10 | 北京师范大学 | A kind of prediction technique and system of answering information |
| CN110476159A (en) * | 2017-03-30 | 2019-11-19 | 日本电气株式会社 | Information processing system, characteristic value illustration method and characteristic value read-me |
| CN110533992A (en) * | 2019-09-02 | 2019-12-03 | 西南大学 | A kind of learning effect prediction technique and system |
| CN110751261A (en) * | 2018-07-23 | 2020-02-04 | 第四范式（北京）技术有限公司 | Training method and system and prediction method and system of neural network model |
| CN111191859A (en) * | 2020-02-27 | 2020-05-22 | 桂林理工大学 | Examination prediction method and system |
| CN111260230A (en) * | 2020-01-19 | 2020-06-09 | 西北大学 | Academic early warning method based on lifting tree model |
| CN111461394A (en) * | 2020-02-24 | 2020-07-28 | 桂林电子科技大学 | Student score prediction method based on deep matrix decomposition |
| CN112116142A (en) * | 2020-09-14 | 2020-12-22 | 桂林电子科技大学 | Student score prediction system and method based on deep learning |
| CN112446558A (en) * | 2021-01-29 | 2021-03-05 | 北京世纪好未来教育科技有限公司 | Model training method, learning result acquisition method, device, equipment and medium |
| CN108241625B (en) * | 2016-12-23 | 2021-10-01 | 科大讯飞股份有限公司 | Method and system for predicting student achievement change trend |
| CN113656749A (en) * | 2021-08-26 | 2021-11-16 | 北京通建泰利特智能系统工程技术有限公司 | Intelligent education management method and system based on big data and readable storage medium |
| CN113705679A (en) * | 2021-08-30 | 2021-11-26 | 北京工业大学 | Student score prediction method based on hypergraph neural network |
| CN113988409A (en) * | 2021-10-27 | 2022-01-28 | 中国海洋大学 | Learning situation prediction method based on multilayer gated neural network |
| CN114372618A (en) * | 2021-12-27 | 2022-04-19 | 北京北明数科信息技术有限公司 | Student score prediction method and system, computer equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1560754A (en) * | 2004-02-25 | 2005-01-05 | 华南理工大学 | Method for presetting condition of rule changing procedure and its device |
| CN1752356A (en) * | 2005-08-11 | 2006-03-29 | 西安理工大学 | Intelligent model constructuring method for analyzing cavity wall rock stability |
| CN104867074A (en) * | 2015-05-15 | 2015-08-26 | 东北大学 | Student comprehensive quality evaluation method based on genetic algorithm optimization BP neural network |
-
2016
- 2016-06-20 CN CN201610443524.7A patent/CN106157204A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1560754A (en) * | 2004-02-25 | 2005-01-05 | 华南理工大学 | Method for presetting condition of rule changing procedure and its device |
| CN1752356A (en) * | 2005-08-11 | 2006-03-29 | 西安理工大学 | Intelligent model constructuring method for analyzing cavity wall rock stability |
| CN104867074A (en) * | 2015-05-15 | 2015-08-26 | 东北大学 | Student comprehensive quality evaluation method based on genetic algorithm optimization BP neural network |
Non-Patent Citations (1)
| Title |
|---|
| 陈勇: "基于遗传神经网络成绩预测的研究与实现", 《现代电子技术》 * |
Cited By (28)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106682768A (en) * | 2016-12-08 | 2017-05-17 | 北京粉笔蓝天科技有限公司 | Prediction method, system, terminal and server of test score |
| CN108241625B (en) * | 2016-12-23 | 2021-10-01 | 科大讯飞股份有限公司 | Method and system for predicting student achievement change trend |
| CN110476159A (en) * | 2017-03-30 | 2019-11-19 | 日本电气株式会社 | Information processing system, characteristic value illustration method and characteristic value read-me |
| CN108280531B (en) * | 2017-07-28 | 2021-07-09 | 淮阴工学院 | Student class score ranking prediction method based on Lasso regression |
| CN108280531A (en) * | 2017-07-28 | 2018-07-13 | 淮阴工学院 | A kind of student class marks sequencing prediction technique returned based on Lasso |
| CN107726632A (en) * | 2017-09-29 | 2018-02-23 | 深圳和而泰智能控制股份有限公司 | Method for heating and controlling and server |
| CN107895033A (en) * | 2017-11-24 | 2018-04-10 | 淮阴工学院 | A kind of method for early warning of student's academic warning system based on machine learning |
| CN108171358A (en) * | 2017-11-27 | 2018-06-15 | 科大讯飞股份有限公司 | Score prediction method and device, storage medium and electronic device |
| CN108171358B (en) * | 2017-11-27 | 2021-10-01 | 科大讯飞股份有限公司 | Score prediction method and device, storage medium and electronic device |
| CN108320045A (en) * | 2017-12-20 | 2018-07-24 | 卓智网络科技有限公司 | Student performance prediction technique and device |
| CN110751261B (en) * | 2018-07-23 | 2024-05-28 | 第四范式（北京）技术有限公司 | Training method and system and prediction method and system for neural network model |
| CN110751261A (en) * | 2018-07-23 | 2020-02-04 | 第四范式（北京）技术有限公司 | Training method and system and prediction method and system of neural network model |
| CN109214579B (en) * | 2018-09-20 | 2020-09-25 | 山东省林业科学研究院 | BP neural network-based saline-alkali soil stability prediction method and system |
| CN109214579A (en) * | 2018-09-20 | 2019-01-15 | 山东省林业科学研究院 | Salt-soda soil stability prediction method and system based on BP neural network |
| 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 |
| CN110223553A (en) * | 2019-05-20 | 2019-09-10 | 北京师范大学 | A kind of prediction technique and system of answering information |
| CN110533992A (en) * | 2019-09-02 | 2019-12-03 | 西南大学 | A kind of learning effect prediction technique and system |
| CN111260230A (en) * | 2020-01-19 | 2020-06-09 | 西北大学 | Academic early warning method based on lifting tree model |
| CN111461394A (en) * | 2020-02-24 | 2020-07-28 | 桂林电子科技大学 | Student score prediction method based on deep matrix decomposition |
| CN111191859A (en) * | 2020-02-27 | 2020-05-22 | 桂林理工大学 | Examination prediction method and system |
| CN112116142A (en) * | 2020-09-14 | 2020-12-22 | 桂林电子科技大学 | Student score prediction system and method based on deep learning |
| CN112446558A (en) * | 2021-01-29 | 2021-03-05 | 北京世纪好未来教育科技有限公司 | Model training method, learning result acquisition method, device, equipment and medium |
| CN113656749A (en) * | 2021-08-26 | 2021-11-16 | 北京通建泰利特智能系统工程技术有限公司 | Intelligent education management method and system based on big data and readable storage medium |
| CN113705679A (en) * | 2021-08-30 | 2021-11-26 | 北京工业大学 | Student score prediction method based on hypergraph neural network |
| CN113705679B (en) * | 2021-08-30 | 2024-05-28 | 北京工业大学 | Student achievement prediction method based on hypergraph neural network |
| CN113988409A (en) * | 2021-10-27 | 2022-01-28 | 中国海洋大学 | Learning situation prediction method based on multilayer gated neural network |
| CN113988409B (en) * | 2021-10-27 | 2024-07-26 | 中国海洋大学 | Learning condition prediction method based on multilayer gate control neural network |
| CN114372618A (en) * | 2021-12-27 | 2022-04-19 | 北京北明数科信息技术有限公司 | Student score prediction method and system, computer equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106157204A (en) | A kind of Forecasting The Scores method and system based on BP neural network model | |
| CN106127634A (en) | A kind of Forecasting The Scores method and system based on model-naive Bayesian | |
| Kahlenberg | From all walks of life: New hope for school integration. | |
| US20140065590A1 (en) | Method, system, and computer-readable recording medium for providing education service based on knowledge units | |
| Zhang | Research on classroom teaching evaluation and instruction system based on GIS mobile terminal | |
| CN113011196A (en) | Concept-enhanced representation and one-way attention-containing subjective question automatic scoring neural network model | |
| CN102663910B (en) | Automatic questions selecting method of examination system on network based on layered genetic algorithm | |
| Kargapolova et al. | Information competences of young people within digitalization of society | |
| Li et al. | [Retracted] Construction of Innovation and Entrepreneurship Platform Based on Deep Learning Algorithm | |
| CN109359190A (en) | A kind of position analysis model construction method based on evaluation object camp | |
| Li et al. | Impacts on Statistics Education in Big Data Era. | |
| An | [Retracted] Based on the Application of Deep Learning in College Education Evaluation | |
| Wang et al. | Mining multi-source campus data: an empirical analysis of student portrait using clustering method | |
| Li | Application of an Internet of things oriented network education platform in English language teaching | |
| Su et al. | A intelligent Gamifying learning recommender system integrated with learning styles and Kelly repertory grid technology | |
| Yang | Evaluation method of English online course teaching effect based on ResNet algorithm | |
| Jianyun | Big data assisted online teaching platform for ideological and political theory course in universities | |
| Hao | Design and application of an intelligent communication platform for aesthetics education for internationalization | |
| Lu et al. | The Construction of College English Teaching Model from the Perspective of Internet‐of‐Things Wireless Communication | |
| Li et al. | Evolution of big-data-enhanced higher education systems | |
| Li | University ideological education quality assessment based on fuzzy comprehensive evaluation | |
| Yuan et al. | Research on the Model of “data-based governance” in Colleges and Universities | |
| Mutovkina | Digital Technologies in the Educational Process and the Effectiveness of Their Use | |
| Shao | Applying PCA and Clustering Algorithm to Analyze the Current Situation of Teaching Ability of Physical Education Teachers in Local Universities | |
| Zhao | Evaluation research on the effectiveness of practical teaching of ideological and political courses in higher vocational colleges under micro-culture |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161123 |
|
| RJ01 | Rejection of invention patent application after publication |