An indicator early warning method and system based on data mining

Abstract

The present invention discloses an indicator early warning method and system based on data mining, the method comprising: obtaining student activity time series data and classifying and quantifying them, establishing a monitoring database; establishing a lasso prediction model according to the time series monitoring data of student quality and literacy indicators in the monitoring database, screening indicators through the lasso prediction model, and determining multiple key indicators; multi-dimensionally dividing the time series monitoring data of key indicators, using the Granger algorithm to train the causal relationship prediction model between key indicators and student quality and literacy evaluation values; predicting the Granger causal scores between different dimensions, different quality and literacy indicators and student quality and literacy evaluation values ​​through the causal relationship prediction model; and early warning of student quality and literacy indicators through the Granger causal scores. The present invention can analyze the causal relationship between different quality and literacy indicators under different dimensions and student quality and literacy evaluation values, and timely perform indicator early warning.

Description

Technical Field

[0001] The present invention belongs to the technical field of data mining, and in particular relates to an indicator early warning method and system based on data mining. Background Art

[0002] Establishing quality and literacy indicators for students is the main means of measuring students' performance in school. Teaching management personnel not only need to pay attention to the single or multiple quality and literacy indicators of a single student, but also need to divide them into units such as class, major, and grade. Through comparisons in different dimensions, the shortcomings in the teaching process can be discovered, and timely alarms can be issued for batch problems, so as to correct the problems that have occurred in past quality and literacy activities and return them to the correct teaching track.

[0003] The traditional quality and literacy monitoring system has a large number of indicators, so it needs to rely on a lot of manual work for statistics and evaluation. In addition to the possibility of human statistical errors, there are also certain requirements for the judges, usually experienced teaching management personnel. If the workload involved in the evaluation is large and multiple people have to be used for evaluation, inconsistent standards may also occur. Summary of the invention

[0004] In view of this, the present invention proposes an indicator early warning method and system based on data mining, which is used to solve the problem that real-time early warning cannot be performed when the setting of students' quality and literacy indicators is unreasonable.

[0005] In a first aspect, the present invention discloses an indicator early warning method based on data mining, the method comprising:

[0006] Obtain the time series data of students' activities, classify and quantify various types of students' quality and literacy activities, obtain the time series monitoring data of students' quality and literacy indicators and the corresponding students' quality and literacy evaluation values, and establish a monitoring database;

[0007] A lasso prediction model is established based on the time series monitoring data of students' quality and literacy indicators, and indicators are screened through the lasso prediction model to determine multiple key indicators;

[0008] Divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions;

[0009] For the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than the preset threshold, the Granger algorithm is used to train the causal relationship prediction model between the key indicators and the evaluation values ​​of students' quality and literacy;

[0010] Predicting the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​through the causal relationship prediction model;

[0011] The Granger causality score is used to provide early warning for students' quality and literacy indicators.

[0012] Preferably, after establishing the monitoring database, the method further includes: preprocessing the monitoring data in the monitoring database, wherein the preprocessing includes quality assessment and data cleaning and standardization.

[0013] Preferably, the expression of the lasso prediction model is:

[0014] Y=X T β+ε

[0015] Where X is the input matrix composed of the time series monitoring data of the student quality and literacy indicators after standardization, β is the parameter matrix of X, Y is the response matrix, and ε is the random error term of the model;

[0016] By constraints The standardized indicator data are screened and the indicator data that do not meet the constraints are eliminated, where λ is the penalty coefficient, which is determined by the K-fold cross-validation method.

[0017] Preferably, the multi-dimensional division of the time series monitoring data of key indicators specifically includes: dividing the time series monitoring data of key indicators into dimensions according to class, grade, and home address area, each dimension including multiple sub-dimensions.

[0018] Preferably, determining the strength of the influence relationship between different dimensions specifically includes:

[0019] Mutual information is used to measure the strength of the influence relationship between different classes, grades or home address areas;

[0020] For the time series monitoring data corresponding to the key indicator of any sub-dimension in each dimension, the time series monitoring data having an influence relationship with the sub-dimension greater than a preset threshold is screened out.

[0021] Preferably, the causal relationship prediction model between the key indicators and the student quality and accomplishment evaluation values ​​trained by the Granger algorithm specifically includes:

[0022] Input the time series monitoring data corresponding to the key indicators of any sub-dimension in a certain dimension to generate sample data for training the causal relationship model;

[0023] Use the Granger algorithm to train the structure and parameters of the causal model;

[0024] A causal relationship prediction model is obtained, which outputs the Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension.

[0025] Preferably, the Granger causal scores between different dimensions, different quality and literacy indicators and student quality and literacy evaluation values ​​are predicted by the causal relationship prediction model specifically including:

[0026] The Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension are predicted by the causal relationship prediction model;

[0027] The Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension are weighted summed to obtain the Granger causal scores between different quality and literacy indicators in different dimensions and students' quality and literacy evaluation values.

[0028] In a second aspect, the present invention discloses an indicator early warning system based on data mining, the system comprising:

[0029] Data acquisition module: used to obtain the time series data of students' activities, classify and quantify the various quality and literacy activities of students, obtain the time series monitoring data of students' quality and literacy indicators and the corresponding students' quality and literacy evaluation values, and establish a monitoring database;

[0030] Indicator screening module: used to establish a lasso prediction model based on the time series monitoring data of students' quality and literacy indicators, screen indicators through the lasso prediction model, and determine multiple key indicators;

[0031] Model training module: used to divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions; for the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than the preset threshold, the Granger algorithm is used to train the causal relationship prediction model between the key indicators and the student quality and literacy evaluation value;

[0032] Causal prediction module: used to predict the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​through the causal relationship prediction model;

[0033] Indicator early warning module: used to provide early warning for students' quality and literacy indicators through the Granger causality score.

[0034] In a third aspect of the present invention, an electronic device is disclosed, comprising: at least one processor, at least one memory, a communication interface and a bus;

[0035] Wherein, the processor, memory, and communication interface communicate with each other via the bus;

[0036] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement the method according to the first aspect of the present invention.

[0037] According to a fourth aspect of the present invention, a computer-readable storage medium is disclosed, wherein the computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to implement the method described in the first aspect of the present invention.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] 1) The present invention first uses the lasso prediction model to preliminarily screen the pre-set student quality and literacy evaluation indicators, eliminates irrelevant indicators, determines multiple key indicators, reduces the data dimension, and improves the calculation efficiency of the student quality and literacy indicator analysis;

[0040] 2) Based on the preliminary screening, the present invention divides the time series monitoring data of key indicators into multiple dimensions, and uses the Granger algorithm to train the causal relationship prediction model between the key indicators and the student quality and literacy evaluation values; the Granger causal scores between different dimensions, different quality and literacy indicators and the student quality and literacy evaluation values ​​are predicted by the causal relationship prediction model. In addition to analyzing the implicit relationship between data of different dimensions, the causal relationship between different quality and literacy indicators under different dimensions and the student quality and literacy evaluation values ​​can be further analyzed. According to the size of the causal relationship, indicator early warning can be carried out, unqualified indicators can be eliminated in time, and accurate indicator screening can be achieved, which can improve the accuracy of student quality and literacy evaluation.

[0041] 3) The present invention analyzes the evaluation indicators of students' quality and literacy through data mining. While realizing analysis and prediction in different dimensions, it also provides real-time warnings for situations that have occurred or may occur, such as unqualified quality and literacy indicators and deviations from the training track, thereby helping teaching management personnel to modify teaching plans, change teaching personnel, or provide data basis for teaching assessment, greatly improving the efficiency of teaching management. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 A schematic diagram of a flow chart of an indicator early warning method based on data mining provided by an embodiment of the present invention;

[0044] Figure 2 A structural diagram of an indicator early warning system based on data mining provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] See also Figure 1 The present invention proposes an indicator early warning method based on data mining, comprising:

[0047] S1. Obtain the time series data of students' activities, classify and quantify various types of students' quality and literacy activities, obtain the time series monitoring data of students' quality and literacy indicators and the corresponding students' quality and literacy evaluation values, and establish a monitoring database;

[0048] Specifically, the student quality literacy indicators include cultural literacy, ability literacy, physical and mental quality, etc., which can be further refined to set cultural literacy indicators such as classroom knowledge mastery, extracurricular knowledge accumulation, cross-disciplinary scientific knowledge accumulation, and self-study initiative, ability literacy indicators such as communication skills, organizational coordination skills, and engineering practice skills, and physical and mental quality indicators such as moral literacy, physical quality, and psychological quality. After obtaining the time-series monitoring data of the student quality literacy indicators, a comprehensive evaluation of the student quality literacy is conducted to obtain the corresponding evaluation value.

[0049] After the monitoring database is established, the monitoring data in the monitoring database is preprocessed, and the preprocessing includes quality assessment and data cleaning and standardization.

[0050] S2. Establish a lasso prediction model based on the time series monitoring data of students’ quality and literacy indicators, screen indicators through the lasso prediction model, and determine multiple key indicators;

[0051] Specifically, the expression of the lasso prediction model is:

[0052] Y=X T β+ε

[0053] Where X is the input matrix composed of the time series monitoring data of the student quality and literacy indicators after standardization, β is the parameter matrix of X, Y is the response matrix, and ε is the random error term of the model;

[0054] By constraints The standardized indicator data is screened, where λ is the penalty coefficient, which is determined by the K-fold cross validation method. The lasso prediction model performs regression analysis to eliminate indicator data that do not meet the constraints, thereby determining multiple key indicators and completing the preliminary indicator screening.

[0055] S3. Divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions;

[0056] Step S3 specifically includes the following sub-steps:

[0057] S31, dividing the time series monitoring data of key indicators into dimensions according to class, grade, and home address area, each dimension including multiple sub-dimensions;

[0058] For example, the class dimension is divided into multiple different class sub-dimensions according to the different classes the students belong to. Similarly, the grade dimension includes multiple different grade sub-dimensions, and the area where the home address is located includes multiple different area sub-dimensions.

[0059] S32. Use mutual information to measure the strength of the influence relationship between different classes, grades or home address areas;

[0060] Specifically, the mutual information between sub-dimensions of different classes, different grades, or different home address areas is calculated pairwise, and the strength of the influence relationship is measured by the size of the mutual information.

[0061] S33. For the time series monitoring data corresponding to the key indicator of any sub-dimension in each dimension, screen out the time series monitoring data whose influence relationship with the sub-dimension is greater than a preset threshold.

[0062] S4. For the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than a preset threshold, the Granger algorithm is used to train a causal relationship prediction model between key indicators and student quality and literacy evaluation values;

[0063] Step S4 specifically includes the following sub-steps:

[0064] S41, inputting the time series monitoring data corresponding to the key indicator of any sub-dimension in a certain dimension, performing feature extraction and generating sample data for training the causal relationship model;

[0065] S42. Use the Granger algorithm to train the structure and parameters of the causal relationship model;

[0066] S43. After the model converges, a causal relationship prediction model is obtained that outputs the Granger causal scores between different sub-dimensions, different indicators and student quality and literacy evaluation values ​​in the same dimension.

[0067] S5. Predicting the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​through the causal relationship prediction model;

[0068] Specifically, the Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension are predicted by the causal relationship prediction model;

[0069] The Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension are weighted summed to obtain the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values.

[0070] S6. Use the Granger causality score to provide early warning for students' quality and literacy indicators.

[0071] Specifically, the Granger causality score describes the causal relationship between different quality literacy indicators and the evaluation value of students' quality literacy in different dimensions. The higher the Granger causality score, the higher the causal relationship, indicating that the corresponding indicator in this dimension is more important for the evaluation of students' quality literacy. If the Granger causality score corresponding to an indicator in a certain dimension is low, the lower the causal relationship, indicating that the corresponding indicator in this dimension is of little value for the evaluation of students' quality literacy. When it is lower than the preset threshold, it means that the indicator is no longer suitable for evaluating students' quality literacy, and the indicator is judged to be unqualified. Therefore, an indicator warning is carried out and the indicator is eliminated to reduce the impact of inappropriate indicators on the accuracy of students' quality literacy evaluation.

[0072] The present invention first uses the lasso prediction model to preliminarily screen the pre-set student quality evaluation indicators, eliminate irrelevant indicators, and determine multiple key indicators; on this basis, the time series monitoring data of the key indicators is divided into multiple dimensions, and the Granger algorithm is used to train the causal relationship prediction model between the key indicators and the student quality evaluation values; the Granger causal scores between different dimensions, different quality indicators and the student quality evaluation values ​​are predicted by the causal relationship prediction model. In addition to analyzing the implicit relationship between data of different dimensions, the causal relationship between different quality indicators and the student quality evaluation values ​​can be further analyzed. According to the size of the causal relationship, indicator warnings can be carried out, unqualified indicators can be eliminated in time, and the accuracy of student quality evaluation can be improved. Please refer to Figure 2 , corresponding to the method embodiment, the present invention also proposes an indicator early warning system based on data mining, the system comprising:

[0073] Data acquisition module 10: used to obtain the time series data of student activities, classify and quantify the various quality and literacy activities of students, obtain the time series monitoring data of student quality and literacy indicators and the corresponding student quality and literacy evaluation values, and establish a monitoring database;

[0074] Indicator screening module 20: used to establish a lasso prediction model based on the time series monitoring data of student quality and literacy indicators, screen indicators through the lasso prediction model, and determine multiple key indicators;

[0075] Model training module 30: used to divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions; for the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than a preset threshold, the Granger algorithm is used to train a causal relationship prediction model between the key indicators and the student quality and literacy evaluation value;

[0076] Causal prediction module 40: used to predict the Granger causal scores between different dimensions, different quality and literacy indicators and the quality and literacy evaluation values ​​of students through the causal relationship prediction model;

[0077] Indicator early warning module 50: used to issue early warning for student quality indicators through the Granger causality score.

[0078] The present invention also discloses an electronic device, comprising: at least one processor, at least one memory, a communication interface and a bus; wherein the processor, memory, and communication interface communicate with each other through the bus; the memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement the aforementioned method of the present invention.

[0079] The present invention also discloses a computer-readable storage medium, wherein the computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to implement all or part of the steps of the method described in the embodiment of the present invention. The storage medium includes: a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk, etc., which can store program codes.

[0080] The system embodiment described above is merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be distributed to multiple network units. A person skilled in the art may select some or all of the modules according to actual needs to achieve the purpose of the solution of this embodiment without creative effort.

[0081] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An indicator early warning method based on data mining, characterized in that: The method comprises: Obtaining time series data of student activities, classifying and quantifying various types of quality and literacy activities of students, obtaining time series monitoring data of student quality and literacy indicators and corresponding student quality and literacy evaluation values, and establishing a monitoring database; the student quality and literacy indicators include cultural literacy indicators, ability literacy indicators, and physical and mental quality indicators; A lasso prediction model is established based on the time series monitoring data of students' quality and literacy indicators, and indicators are screened through the lasso prediction model to determine multiple key indicators; Divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions; For the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than the preset threshold, the Granger algorithm is used to train the causal relationship prediction model between the key indicators and the evaluation values ​​of students' quality and literacy; Predicting the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​through the causal relationship prediction model; The Granger causality score is used to provide early warning of students' quality and literacy indicators; After the monitoring database is established, the method further includes: preprocessing the monitoring data in the monitoring database, wherein the preprocessing includes quality assessment and data cleaning and standardization processing; The expression of the lasso prediction model is: Y=X T b+e Where X is the input matrix composed of the time series monitoring data of the student quality and literacy indicators after standardization, β is the parameter matrix of X, Y is the response matrix, and ε is the random error term of the model; By constraints The standardized indicator data are screened to eliminate the indicator data that do not meet the constraints, where λ is the penalty coefficient, which is determined by the K-fold cross-validation method; The multi-dimensional division of the time series monitoring data of key indicators specifically includes: dividing the time series monitoring data of key indicators into dimensions according to class, grade, and home address area, each dimension including multiple sub-dimensions; Determining the strength of the influence relationship between different dimensions specifically includes: Mutual information is used to measure the strength of the influence relationship between different classes, grades or home address areas; For the time series monitoring data corresponding to the key indicator of any sub-dimension in each dimension, screen out the time series monitoring data whose influence relationship with the sub-dimension is greater than a preset threshold; The causal relationship prediction model between the key indicators and the student quality evaluation values ​​trained by the Granger algorithm specifically includes: Input the time series monitoring data corresponding to the key indicators of any sub-dimension in a certain dimension to generate sample data for training the causal relationship model; Use the Granger algorithm to train the structure and parameters of the causal model; A causal relationship prediction model is obtained to output the Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension; The Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​predicted by the causal relationship prediction model specifically include: Predicting the Granger causal scores between different sub-dimensions, different indicators and student quality evaluation values ​​in the same dimension in new student activity time series data through the causal relationship prediction model; The Granger causal scores between different sub-dimensions, different indicators and students' quality and literacy evaluation values ​​in the same dimension are weighted summed to obtain the Granger causal scores between different quality and literacy indicators in different dimensions and students' quality and literacy evaluation values.

2. An indicator early warning system based on data mining, which uses the indicator early warning method based on data mining according to claim 1, characterized in that: The system comprises: Data acquisition module: used to obtain the time series data of students' activities, classify and quantify the various quality and literacy activities of students, obtain the time series monitoring data of students' quality and literacy indicators and the corresponding students' quality and literacy evaluation values, and establish a monitoring database; Indicator screening module: used to establish a lasso prediction model based on the time series monitoring data of students' quality and literacy indicators, screen indicators through the lasso prediction model, and determine multiple key indicators; Model training module: used to divide the time series monitoring data of key indicators into multiple dimensions to determine the strength of the influence relationship between different dimensions; for the time series monitoring data of each dimension and the corresponding time series monitoring data with an influence relationship greater than the preset threshold, the Granger algorithm is used to train the causal relationship prediction model between the key indicators and the student quality and literacy evaluation value; Causal prediction module: used to predict the Granger causal scores between different dimensions, different quality and literacy indicators and students' quality and literacy evaluation values ​​through the causal relationship prediction model; Indicator early warning module: used to provide early warning for students' quality and literacy indicators through the Granger causality score.

3. An electronic device, characterized in that: include: at least one processor, at least one memory, a communication interface, and a bus; Wherein, the processor, memory, and communication interface communicate with each other via the bus; The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to implement the method according to claim 1.

4. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions that cause the computer to implement the method of claim 1 .

Images