Dynamic early warning system of College Students’ target course performance based on improved Apriori algorithm

Abstract

The early warning system of College Students’ target course achievement is an important part of the educational administration system in Colleges and universities. This paper proposes to use some techniques of association principle to mine a large amount of data in the performance system to a certain extent, and obtain available rules from the data. Based on the characteristics and shortcomings of Apriori algorithm, an improved Apriori is proposed. The algorithm can process and mine the data in the early warning system of College Students’ scores, and finally obtain the management principles, thus forming an effective early warning for the course learning. In order to promote the improvement of students’ academic performance and achieve the ultimate goal of cultivating excellent talents in Colleges and universities.

Keywords

Improved Apriori algorithm college students target course achievement early warning system

1. Introduction

In recent years, some colleges and universities have changed the previous single final examination mode to the phased examination mode, that is, the process assessment. Its basic purpose is to make students pay attention to it all the time at the level of knowledge, so as to avoid students’ failure in a certain course after the end of a semester. At this time, it is too late to wake up. It is difficult to achieve the goal of mastering the knowledge of a certain course only by means of make-up examination and review. College Students’ learning early warning system is a software system based on the process performance assessment method implemented by some colleges and universities in China. The system is an upgrade of the student management system applied in earlier colleges and universities in terms of business processing. The system is designed around how to better use information technology and network platform to supervise students’ learning quality. On the basis of the phased examination business process proposed by the process assessment method, in terms of function, it retains the comprehensive management of students’ scores possessed by the earlier student management system, and proposes the function of in-depth statistics of students’ periodic performance, and finally achieves the purpose of learning early warning [1].

Apriori algorithm is a classic algorithm for mining association rules. The basic idea of Apriori algorithm is as follows: Firstly, frequent item sets are mined out from the transaction database. Compared with the minimum support threshold, all item sets whose support is greater than this threshold belonging to frequent item sets. Then there is the generation of association rules, based on frequent item sets. In China, educational informationization construction starts relatively late. With the gradual expansion of college enrollment scale, the number of college students is increasing day by day. In the face of a huge student group, how to efficiently manage all aspects of students makes the development and application of educational software system gradually put on the agenda. First of all, the student management software is a typical student management system, it takes the student’s basic personal information management as the main business, to save the student information in the form of personal files. With the development of computer software and the need of university teaching management, student achievement management system has been widely used in universities. With the reform of higher education system, the student achievement management system has gradually increased the functions of credit management and elective course management. However, there is still a big gap between China and foreign countries. The first is the gap between education systems. The development of a software system usually depends on the business in which the system participates. When the education system falls behind, software systems can’t make big breakthroughs in function. In this regard, the relevant departments of our government have to quickly establish the corresponding education information early warning system and rapid response mechanism, improve the emergency management system, and comprehensively improve the education management ability. In our country, educational informationization construction starts relatively late. With the gradual expansion of college enrollment scale, the number of college students is increasing day by day. In the face of a huge student group, how to efficiently manage all aspects of students makes the development and application of educational software system gradually put on the agenda. First of all, the student management software is a typical student management system, which takes the management of students’ basic personal information as its main business and saves students’ information in the form of personal files [2]. With the development of computer software and the need of university teaching management, student achievement management system has been widely used in universities. With the reform of higher education system, the student achievement management system has gradually increased the functions of credit management and elective course management. However, there is still a big gap between China and foreign countries. The first is the gap between education systems. The development of a software system usually depends on the business in which the system participates. In the case of the backward education system, the software system cannot achieve a major breakthrough in function [3].

Crisis warning can not be calculated and evaluated scientifically. It has been regarded as a subject of great social value, educational theory and practice. The proposal of data mining technology and its application in the field of crisis management and early warning provide strong support, making the analysis process of crisis warning more smooth and the prediction results more accurate [4]. The mining of association rules is one of the important contents of data mining. It shows the association relationship between data, and it is easy to understand. It mainly finds the association relationship between item sets or attributes that meet the minimum support and confidence from transaction database or relational database [5]. Apriori algorithm is an algorithm of mining frequent item sets using Boolean association rules with layer by layer search iteration. This paper introduces the classic Apriori algorithm in data mining technology and its main defects, and puts forward the improved Apriori algorithm. It analyzes the relationship between College Students’ scores by using the improved Apriori algorithm, processes and mines the data in the early warning system of College Students’ scores, and finally obtains the management principles, thus forming an effective early warning for course learning [6].

2. Improved Apriori algorithm

2.1 Apriori algorithm and its shortcomings

Apriori algorithm has been regarded as the most widely used association rule mining algorithm. Generally speaking, when we understand Apriori algorithm, we think that the main idea of Apriori algorithm is to use probability knowledge in mathematical statistics to mine Boolean association rules of data to a certain extent [7]. When the Apriori algorithm is used to continue the calculation, the original structure data is first layered, and then the data of each layer are processed by the same method. Generally speaking, when using Apriori algorithm, we need to follow two steps: The first step is the mining process, that is, mining all the frequent item sets from the database; The second step is based on the frequent item sets mined in the first step, and finally obtain the association rules that meet the conditions. As a classical Boolean association rule algorithm, Apriori algorithm has a relatively complex process. It needs to search the database layer by layer for many times, and iterate the frequent sets in the database with the help of prior knowledge of frequent item sets until there is no frequent set [8]. However, this algorithm still has some defects. Its defects are mainly reflected in several aspects: One is to increase the computational complexity of the algorithm. Generally speaking, Apriori algorithm is mainly affected by four factors: support threshold, number of transactions, number of items and average width of transactions. In Apriori algorithm, the decrease of support threshold will lead to the increase of frequent item sets to a certain extent, which will increase the difficulty and complexity of calculation. The reason is that when the number of frequent item sets increases, the number of candidate item sets counted increases, and the maximum length of frequent item sets increases correspondingly, resulting in the increase of the number of scans required by the algorithm [9]. Second, the number of transactions. When using Apriori algorithm to calculate the support of item sets, it is necessary to scan the data sets repeatedly. In this case, the number of transactions increases, and the corresponding running time increases. Third, the candidate pruning methods increase the computational complexity. In addition, the shortcomings of Apriori algorithm are also reflected in the increase of computation and I/O overhead.

2.2 Improved Apriori algorithm

Under the network environment, the application of advanced modern information technology and network technology constantly promotes the process of information construction in Colleges and universities, and provides a good technical guarantee for the management and operation of university educational administration system. At the same time, with the advancement of information construction in Colleges and universities, the data generated in the educational administration system is also increasing rapidly. In the practical application of educational administration system, users only have basic operation authority, and the existence value and association relationship of all kinds of data in the system can not be effectively mined. When analyzing the relevance of data and mining its value, the traditional Apriori algorithm itself has some defects. Therefore, it is necessary to improve Apriori algorithm.

In the process of discovering frequent item sets, Apriori algorithm will generate a large number of candidate item sets, and with the scanning of data sets again and again, resulting in a large time consumption of I/O [10]. According to the improved algorithm, the items of interest are set. After scanning the database for the first time, a new database is obtained. At the same time, the judgment data set is added, and the irrelevant and meaningless candidate item set is deleted, which reduces the time consumption cost. The operation flow of improved Apriori algorithm is shown in Fig. 1.

Figure 1.

Improved Apriori algorithm flow.

The specific process of Apriori algorithm is as follows:

(1)

Frequent 1-item sets were found. First scan the new data set and find frequent 1-item sets according to the minimum support set by the user.

(2)

Add judgment data set. It is helpful to generate pruning of candidate k-item sets and reduce the cost of time consumption.

(3)

Frequent k-item sets are found. All frequent k-item sets are generated according to the support measures, and then meaningful rules are obtained according to the minimum confidence measure set by users.

2.3 Definition of improved Apriori algorithm

Definition 1 student curriculum item set (desert_ Itemset): Association rule problems based on students’ courses can be expressed as follows:

Let the item set Di $=$ {di1, di2, di3, did} is the collection of all the attribute data in the student course, and D represents the number of attribute data in the course.

Definition 2 student course Affairs (desert_ Transaction): DT $=$ {dt1, dt2, dt3, … DTN} represents the attribute transaction set owned by the student curriculum, and N represents the number of research and analysis target courses. Each transaction DTN is a subset of attribute set Di, namely $dt_{n}\subseteq DI$ .

Definition 3 student course support (desert)_ Support: The probability that DTA and DTB occur simultaneously on sheep in desert grassland is called student course support. That is, the probability of two attribute items occurring at the same time in students’ courses, as shown in Eq. (1):

$\displaystyle\textit{DSupport}({DT_{a}\Rightarrow DT_{b}})=P({DT_{a}\cup DT_{b% }})=\frac{\textit{Support\_count}({DT_{a}\cup DT_{b}})}{\textit{Total\_count}(% {DT})}$ (1)

In Eq. (1), $\textit{Support\_count}({DT_{a}\cup DT_{b}})$ represents the number of DTA and DTB simultaneously occurring in all transactions, and $\textit{Total\_count}({DT})$ represents the number of all student course transactions [11].

Definition 4 student course confidence (desert_ Confidence): under the condition that the student curriculum attribute set DTA occurs, the probability of occurrence of student curriculum attribute set DTB is called the confidence degree of student curriculum association rule, and the calculation method is shown in Eq. (2):

$\displaystyle\textit{DConfidence}({DT_{a}\Rightarrow DT_{b}})=P(DT_{a}|DT_{b})% =\frac{\textit{Support\_count}({DT_{a}\cup DT_{b}})}{\textit{Support\_count}({% DT_{a}})}$ (2)

The expression form of association rules in definition 5 is: DTA $\to$ DTB, where DTA and DTB represent the antecedent and subsequent items of the rule respectively, where DTA $\subset$ I, DTB $\subset$ I and $DT_{a}\cap DT_{b}=\emptyset$ AA. E measurement of rules is not less than the minimum confidence and minimum support.

2.4 Improve the code form of Apriori algorithm

The code form of improved Apriori algorithm to generate frequent 1-item sets is shown in Table 1.

Table 1
Code form of improving Apriori frequent 1-item sets generation

Algorithm 1: Generate frequent 1-item sets
Input: transaction data set
Output: Frequent 1-Item Set
(1) for ti $\in$ D
(2) for i $\in$ ti
(3) Output $<$ c1, 1 $>$
(4) end for
(5) end for
(6) for ti $\in$ D
(7) if c1 $\in$ ti
(8) c1_count $++$
(9) end if
(10) end if
(11) support $=$ c1_count/total_count
(12) if (support) $\geqslant$ min_sup)
(13) Output $<$ L1, support $>$
(14) end if

In the improved Apriori algorithm, frequent 1-item sets L1 and candidate 2-item sets C2 are generated. According to the characteristics of students’ curriculum data with various attributes, attribute reduction is conducted for candidate 2-item sets in the process of generating candidate 2-item sets (that is, data with the same attribute cannot constitute candidate item sets), thus generating frequent 2-item sets [12, 13]. The algorithm for generating frequent k-item sets is shown in Table 2.

Table 2

Code forms for generating frequent k-item sets

Algorithm 1: Find frequent

k

-item sets

Frequent (k-1)-item set

L_{k-1}

(

k\geqslant

2) and judgment data set shuxing

=

[A, B, C, D]

Output: frequent

k

-items set

L_{k}

(1) read, shuxing $=$ [A, B, C, D]

(2) for i to shuxing_count do

(3) if can $\in$ shuxing [i]

(4) delect can

(4) else

(6) Output can

\to C_{k}

(7) end for

(8) for ti $\in$ D

(9) c1 $=$ subset (

C_{k},T_{i}

)

(10) end if

(11) support $=$ count/total_count

(12) if (support)

\geqslant

min_sup)

(13) output $<L_{k}$ , support $>$

(14) end

The students’ courses are classified into several sub databases, each sub database contains the scores of the courses covered by their subordinate classifications. When mining frequent item sets of each sub database, the vertical data format is mainly used. On the basis of mining association rules separately, the rule results obtained are integrated.

The improved Apriori algorithm is described as follows: First, the database is compressed by preprocessing. The student achievement system contains a large number of student achievement data. By preprocessing the score database, it is compressed into the performance transaction database. The second is to classify students’ courses into three sub databases according to the differences of course contents. Each sub database contains the course scores of all courses in the corresponding category. Third, on the basis of the second step classification, the corresponding three groups of frequent item sets are generated. (1) Firstly, the database is scanned once, and then it is rebuilt after disorder. After adjusting the transaction project database, it is established as the project transaction database. The project data should be fully counted to ensure that there is no omission. Some item sets that do not reach the minimum support should be deleted. (2) Using the connection function of Apriori algorithm, the intersection of items in the project transaction database is calculated, and then the new data is reconstructed into a new database. Finally, the n-item frequent itemset is obtained. The fourth is to fully integrate the frequent item sets in all sub databases to obtain the confidence level of each association rule. According to the minimum confidence threshold, all strong association rules are obtained to form a global candidate frequent item set. Finally, these data are updated in the system.

3. Student performance warning model based on improved Apriori algorithm

Based on the improved Apriori algorithm, the college students’ learning early warning system should be based on the actual business of colleges and universities, and deeply study the implementation rules of process learning management in process learning management and process module examination, that is, referring to the actual business of colleges and universities in this teaching arrangement, and making full use of the business operation advantages of the software system, a reasonable business process is designed. At this stage, the business arrangement of the system starts from the teachers of various subjects as the main object of use, so that teachers can quickly input the data of students’ attendance, classroom performance and course module scores through the system, and allows teachers to modify the above scores. These three aspects of performance management interface should be based on the assessment of different emphasis and different. The traditional teaching manual or excel table statistics and other data organization means can be reasonably transformed into the operation mode through web page control (As shown in Fig. 2).

Figure 2.

Dynamic warning system.

3.1 Score statistics

Results statistics are usually completed by the head teacher or Counselor of each class. In the past operation process, the statistical work reflects the problem of large amount of data and difficulty in statistics. The score statistics is divided into the stage score statistics and the final total score statistics. The stage score is to sum the attendance score, classroom performance score and course module score of each course module, and calculate the 100-point system score. The average score of each stage of the course module is worth the total score of a certain student at the end of the term. This score statistics system makes students not only reflect their usual performance, but also have their final total score for reference. The statistical business of stage performance plays a major role in the early warning process of students’ learning. Through the analysis of two or more adjacent stages of performance, we can often see the extent of students’ progress or backwardness. Through the analysis of attendance and classroom performance, we can see the impact of changes in students’ thinking or behavior on learning quality, so as to better warn students’ learning situation. The early warning process in the real business is that the head teacher talks to the students who have obvious differences between the two stages. However, due to the long period of data statistics and inaccurate analysis of data stage, the work of early warning for students’ learning is not enough.

College Students’ learning early warning system based on improved Apriori algorithm should increase the comprehensive statistical function of the score data in business processing. The core function of the system is the early warning operation of students’ learning. The fast and good early warning business processing directly affects the quality of the system’s core functions, and indirectly affects the improvement of students’ learning quality. Therefore, in the work of stage performance statistics, the system should be based on the attendance score, classroom performance score, course module score three aspects of statistics, also should include the comprehensive calculation of the stage performance. On the basis of the above statistical results, we should judge the stage of the phased examination at the same time. If it is the assessment of the first stage, there is no need to conduct in-depth analysis of the data, and only need to segment the current data to determine the scores of students who fail, pass, good and excellent. And the current statistical results for the second. In the next stage, it is necessary to conduct in-depth analysis of the data, including the difference analysis of the two attendance results, the difference analysis of the two classroom performance scores, and the difference analysis of the two course module examination scores, so as to judge the students’ learning status in the two stages, and then give early warning and prompt to the students who are lagging behind and praise the students who have made great progress in their learning. With the increase of the number of periodic examinations, students’ learning state will be reflected more intuitively through the data, which is also very conducive to the accurate assessment of students’ learning and then early warning or praise.

3.2 Performance warning

Performance early warning is the core business of College Students’ learning management, and it is also a new measure in teaching research and implementation to improve students’ learning quality. The whole process of performance early warning is completed by the head teacher. In the process of implementation, the teacher in charge of the class will calculate the student’s periodic performance in writing or through Excel, so as to find the students who have made obvious progress in the class or whose performance has declined. Through talking with the backward students themselves, they can guide the students to strengthen their learning level in the next stage. In the actual operation of the performance early warning business, we have completed the learning management with the guidance of the head teacher as the auxiliary, and the improvement of students’ academic performance as the main way.

The system provides an operation page for teachers to view students’ scores. Students’ scores will be presented in a three-dimensional way of charts and numbers. Teachers can log in to the system by inputting the teaching staff number and password to manage the performance warning interface. In the business, the system will add a number of ways to view students’ scores and analyze the results of data. The early warning students will carry out manual check, and the early warning operation will be realized after confirmation. The early warning function not only retains the previous teachers’ direct reminding business for students, but also adds the business operation of SMS or email notification to parents. The system provides query methods such as stage time screening, score category screening, excellent students, backward students, and reasons for differences.

3.3 Functional requirements analysis

The design and implementation process of the software system requires the overall analysis and detailed analysis of the functions of the system. Through the analysis of each function module, the system can be coded and implemented smoothly. The system can be divided into process learning management function, learning situation analysis function, learning early warning function and system management function.

3.3.1 Process learning management function

The process learning management function belongs to the system frequently used function. The function module is further divided into attendance management function, classroom performance management function, course module score management function and process basic setting function.

(1) Attendance management function

The attendance management function realizes the management of students’ daily attendance in class, and is responsible for the statistics and maintenance of students’ attendance. Attendance management is further divided into late, early leave, absenteeism, sick leave and personal leave. Attendance management is used on the premise that the basic data information of the system has been imported. The teacher can find the class student information of the current course by selecting the teaching information, and record the attendance status of no students by checking with the mouse. In order to speed up the management of teachers in this work, the default setting of the system is that the attendance status of all students in the class is normal. Teachers only need to view and record the bad attendance of individual students. The system provides teachers with the function of adding students’ attendance record after class to the classroom without computer.

(2) Classroom performance management function

The main purpose of classroom performance management function is to realize the management function of students’ listening state, to achieve the purpose of restricting students’ classroom behavior and ensuring students have good classroom listening state. The performance of students in class is usually reflected in sleeping, chatting, playing mobile phones, reading novels and so on. The instructor will decide whether to use the system to record based on the severity of student performance. The system provides a quick check function in the classroom performance management. The teacher can make a comparative selection according to the class performance options listed in the function page. The teacher can also add some other students’ performance behavior in the classroom by customizing the content.

(3) Course module score management function

The course module is the result of the further detailed planning of each semester’s curriculum, which is also the teaching content reform scheme to realize the effective supervision of students’ learning process. The operator of this function is the teacher and the object of the function is the course. Through the use of this function, the module division, class hour statistics and score percentage management of a course are realized. It can also achieve a preliminary summary of the overall performance of students in a teaching module.

3.3.2 Learning situation analysis function

The function of learning situation analysis includes attendance statistics, classroom performance statistics and comprehensive statistics. The above three sub functions realize the overall grasp of students’ learning situation and play a key role in the early warning of students’ learning situation.

(1) Attendance statistics function student attendance statistics function realizes the management of students’ daily attendance, and completes the function of recording students’ attendance, lateness, early leave and sick leave. Through this function, teachers can add, modify, query and delete data. Attendance statistics through the web page as the operating platform, in order to make the management of teachers to design, reference to the business logic of the function, according to the default all students have been normal attendance angle design. It is functionally acceptable asp.net. The check box control is provided for the overall layout, so that teachers only need to select the students who are not in attendance or have bad attendance such as being late. Through the page, you can make a basic summary of each attendance, so that teachers can intuitively understand the overall attendance of students in the current course. After the end of the course, the class teacher can also log in the system to query the attendance of a course according to the time end or different subjects.

(2) Statistical function of classroom performance

The statistical function of classroom performance can record and count the students’ Misconduct in class, such as whether they sleep, whether they chat, whether they play teaching, and so on. The design purpose of the functional module is to understand the overall situation of students’ classroom learning. In the interface design, the form of drop-down list box is used, and the management can be completed only by the list selection operation of a certain student’s behavior misconduct. At the end of the class, the operation page will also be a comprehensive summary of the overall performance of students in this class. After class, the head teacher can view the overall classroom performance of the students in a certain period of time or a certain course by logging in the system.

(3) Course module score management function

The score management function of course module is used to count the examination results, attendance and class performance of a teaching module of a course. The results of these three aspects will be taken as the final score of a teaching module. The score of the course module reflects the students’ learning status at a certain stage, which will be used as an important reference for the operation of students’ learning warning. The results of the course module allow the teacher in charge of the class to query the learning situation of the course module comprehensively.

3.3.3 Learning early warning function

Learning early warning function is the core function of the system. Its function is to query and remind students of their learning situation, so as to ensure the improvement of students’ learning quality. In the function design, it includes three aspects of early warning operation: attendance warning, classroom performance warning and academic performance warning.

The operation of learning early warning function is based on the data of students’ attendance, classroom performance and academic performance. The function forms an intuitive analysis of a student’s comprehensive performance of a certain course through in-depth data mining. It can timely remind students by means of email and SMS.

This function also introduces the function of early warning and notification to the parents of students, so as to realize the management mechanism of parents and schools’ joint supervision in business. In terms of technology, the comprehensive statistics and early warning functions of comprehensive query window and chart feedback are designed.

(1) Attendance warning function

The early warning function of students’ attendance is to reflect the students’ learning situation from the students’ attendance status and indirectly understand the students’ learning attitude. Attendance is a necessary condition for students to learn a course well. Only by ensuring a good attendance state can we further learn a certain course. In terms of function, it is necessary to conduct statistical operation on the attendance data to form a curve chart of students’ attendance status at a certain stage of a course. The students’ attendance status can be effectively fed back according to the time sequence and course sequence through the form of chart response data.

(2) Early warning function of classroom performance

Classroom performance is the intuitive feedback of whether students can listen to the class carefully. This function can complete the comprehensive statistics and summary operation of the teacher’s record data, and needs to realize the classification and summary of students’ classroom performance. You can query according to the class behavior misconduct, for example: To find out the information of students who often sleep in a certain course. The classification query realized by this function can complete the early warning prompt for some aspect of students.

(3) Stage performance warning

The performance early warning in the student learning early warning system is the key factor to achieve the early warning effect. Attendance early warning and classroom performance warning are reflected in the management of students’ learning attitude, and the level of achievement is the main reference data which ultimately reflects the students’ learning status. The purpose of performance warning is to warn and inform the students whose academic performance is not qualified, so as to achieve the management effect of informing in advance and indirectly promoting the students to study hard. It should have the following features:

1)
The warning time should be before the final examination, and the earlier the warning time, the better.
2)
The score control of score warning should be reasonable and scientific, that is to say, the division of score warning line should be reasonable.
3)
The object of early warning can be extended to parents except students.
4)
The means of early warning should be informationized.

The stage grade early warning function needs to realize the learning warning function after the end of a certain course module. This function is a comprehensive reminder and warning for students’ learning situation at a certain stage. Through the teacher in charge of the class to check the final results of the course module, to screen out students with early warning information, through email, SMS notification and other means to remind students. The warning content sent by the students can be automatically analyzed by the system, and it also needs to support the manual input operation of the teacher. The function also needs to remind the parents of a student. The early warning function of students needs to support the early warning operation in different time periods in terms of time, the early warning operation for students with poor learning in a certain subject, the classification warning operation for attendance, classroom performance and academic performance in categories, the early warning operation for students themselves and their parents in identity, and email in means Early warning sending and SMS sending. In terms of function design, the flexibility, comprehensiveness, real-time and accuracy of function and data processing should be guaranteed.
4. System implementation and testing

4.1 Implementation of data operation technology

The implementation of data operation depends on the design of the underlying data operation of the system, and the data processing is carried out through the data layer. Application in technology ado.net. The technology completes the whole process of data operation, completes the connection operation to the existing database through the database connection object, judges the status of the connection object and opens it [14]. Create database command object, add, delete and modify data. Using the data set object and making full use of the advantage of the data processing speed of the data set object to complete the comprehensive operation of the data. The system creates a data processing layer named Dal, declares an abstract class named sqlhelp, saves database connection strings through attributes, and creates executemymeth () method [15], getdata () and other methods to implement data operation [16].

4.2 Realization of login operation

The implementation of the login operation depends on ado.net. The data layer design of login function needs to call the data query method getdata (), and the entity layer needs to create entity classes Teachers, in addition to the inherent fields and attributes of the login user (teacher), the login method is required to support the login operation process. Therefore, the business layer creates the behavior and rule processing method of the business layer. The business layer calls the getdata () method of the data layer, and the business method login () depends on the entity layer [17]. The representation layer is Login.aspx. Through this page, the user name, password and verification code input box are designed, and the user’s login operation behavior is triggered by clicking the button’s click event. Therefore, the processing flow of login operation is Login.aspx. The page calls the teachers entity class, the entity class calls the login () login method, and adds the login query statement. The login method calls the get data () data query method. The implementation of the presentation layer needs to carry out basic UI design, design the interface that meets the overall needs of the system, and set up three text boxes. At the client side, it needs to verify the validity and correctness of the data values of the three text boxes. The validity verification is reflected in the normal letters, numbers, underscores, etc. the illegal characters include @, ’,’, etc. The correctness verification needs to compare the input character content with the database query results, and judge the number of query records by using the statement select count (*) from users where username =’ and PWD =’, when the query result value is greater than 0, the user exists, otherwise the user does not exist. The data query operation needs to be completed through the data layer getdata () method.

The main purpose of system testing is to find the defects in the software system, and test all aspects of the software system through technical means. This system mainly adopts load test and function test. With the development of web technology and its application scope, manual testing of Web site can not meet the actual operation needs of web system. It is necessary to understand and master the web through testing tools. The system can withstand the running environment, to avoid the risk of system crash or paralysis when multi-user concurrent access and operation. The function test mainly studies the standardization and correctness of each function module in the business operation, so as to avoid correcting the functional defects found in time. The data processing part of login operation adopts jQuery data submission scheme, that is, the post method in jQuery technology is used to statically submit the user name, password and verification code information to the business layer for processing. This technology also belongs to the asynchronous transmission operation in AJAX technology, and the processing results of business layer are returned to the landing page at the front desk According to the returned result, “user is logging in …” And “login failure prompt”, through the application of this technology, the login process will not let the page and server produce the whole page loading process.

4.3 Realization of learning early warning

The realization of learning early warning function is to reflect the students’ periodic learning changes through the way of curve graph. In terms of technology, it mainly adopts JavaScript framework based on jQuery and high charts [18]. Through the curve, we can intuitively and dynamically reflect the students’ learning situation, and at the same time, it also plays a guiding role in predicting the trend of students’ future learning state.

(1) Prepare for learning early warning

The implementation of the function module needs to be imported first highcharts.js. And jquery-1.8.3.js file. The code is as follows:

<scripttype="text/javascript"src="http://zhaowenbinmail.blog.163.com/blog/ js/jquery-1.8.3.min.js"></script> <script src="http://zhaowenbinmail.blog.163.com/blog/js/highcharts.js"></script>

Secondly, the div element needs to be added to the web page to draw the graph based on this div element. The code is as follows:

Finally, the business logic operation of curve drawing is realized by writing new JavaScript code. The specific process is to take the module time information as the abscissa of the chart, take the module score as the vertical coordinate, and establish the horizontal and vertical coordinate information through X and Y attributes. The code is as follows:

X: { title: { enabled: true, text: ’Module assessment time point’ }, max:8, min:0, tickPixelInterval:60 } Y: { title: { text: ’Grade value’ }, plotLines: [{ value: 1, width: 2, color: ’#717171’ }] }

(2) Get background data

In the web page, the method of getbackdata () is created to obtain the background data. The interaction between the foreground and the background is completed by using JavaScript development packages such as jQuery and JSON. The foreground page data is submitted to the application file by post “Default.aspx/geDataTableByAjax”. Through the data layer processing of C code, the data is transformed into JSON. The format is returned to the foreground callback function. If the data is obtained successfully, the JSON data is converted into the result set in the form of table, and the array element values are stored in the coordinate points of the data chart once.

Getbackdata(){ $.ajax({ type:"Post", url:"Default.aspx/geDataTableByAjax", contentType:"application/json; charset=utf-8", dataType:"json", success:function (data){ data= jQuery.parseJSON(data.d); //JSON Convert to Table form againï¼› varseries= chart.series[0]; series.addPoint([data[0]["x"],data[0]["y"]],false,true); chart.redraw(); }, //If an exception occurs error:function (mes){ alert(mes + "Data acquisition failed!"); } }); }

The loading event in jQuery is used to complete the event processing of web page loading. In order to make the data chart display data dynamically in real time, the time processing function in JavaScript is called to realize the process of calling data once every 2 seconds.

$(document).ready(function() { window.setInterval(Getbackdata,2000); }); });

4.4 System test plan

The system testing work will be carried out from two aspects: load test and function test. The load test research work introduces the LoadRunner test tool, which can simulate the running status of the system when a large number of users are concurrent. The tool includes Vugen of pressure script and background monitoring controller. Through the analysis of the final data to master the system test results. Through the use of Vugen function, we can simulate ten thousand level users to ensure the establishment of multi session objects. The controller is used to formulate multi-user operation scheme and simulate the events of multi-user performing the same operation [19], such as multi-user login at the same time, multi-user query a certain data at the same time. Through analysis, the data generated in the above processing process are analyzed and summarized. Refer to Table 3 for test data.

Table 3
System load test results

Number of clients	Average response time	Server CPU usage/%	Server CPU usage
500	0.0903	10	No
1000	2.109	10	No
1500	2.172	12	No
2000	2.210	12	No
2500	2.371	13	No

In order to verify the accuracy of the dynamic early warning system of target course performance, this paper compares the designed early warning system with the early warning system using the methods of literature [7] and document [8]. The results are shown in Fig. 3.

Figure 3.

Comparison of response time of early warning system with different algorithms.

As can be seen from Fig. 3, the response time of the early warning system using the improved algorithm is much shorter than that of the other two algorithms, which shows that the response time of the early warning system based on the improved algorithm is faster, the system has better performance in terms of network throughput and memory utilization, and supports concurrent access and concurrent operation of at least 2 500 people.

In order to verify the accuracy of the dynamic early warning system for the performance of the target course, a comparative experiment was carried out between the method in this paper and the reference [7] and the reference [8]. The results are shown in Fig. 3.

Figure 4.

Comparison of the error of early warning transmission by different methods.

As can be seen from Fig. 4, the transmission error of the proposed method to the early warning system is less than that of the other two algorithms, indicating that the proposed method has strong applicability and higher accuracy.

In the aspect of function test, the four main function modules and sub function modules involved in the system are tested successively. The message passing and business process among UI layer, business layer, entity layer and data layer of web system are tested respectively. Through repairing the system bugs found and summarized, the function test scheme will adopt the equivalent class input test method. Row test and equivalence class test is to divide the data to be input into several data sets according to different categories, select the representative data of each set for input test, and judge whether the output results meet the expected requirements. The equivalent class test cases are designed for each function operation, and the function test plan is described. For example, the test case of login function is described as follows: The test function is login, and the input equivalence classes are wrong account information, correct account information, and account information with special characters. The expected output results are login failure and prompt information. After successful login, you can jump to the main interface of the system, and login failure will prompt illegal login information.

In addition, due to the web system mode, the system also carries out link test and cookie test. It is introduced as follows:

(1) Link test, link is a main feature of web application system, it is the main means to switch between pages and guide users to some pages without knowing the address. Link testing can be divided into three aspects. First, test whether all links are linked to the linked page as indicated; Secondly, test the form of cookies stored on the client computer, which can be used to create dynamic and custom pages or store login information. If the web application system uses cookies, you must check whether the cookies work properly. The content of the test can include whether the cookies work or not and whether they are saved according to the scheduled time.

Finally, ensure that there is no isolated page on the web application system. The so-called isolated page refers to that there is no link to the page, only know the correct URL address can access. The link test of this system uses the SEO tools under the Internet environment for statistics and analysis. By inputting the domain name of the site, the total number of links, positive and negative links, dead links and other data of the system can be analyzed.

(2) Cookies test, cookies are usually used to store user information. When a user uses cookies to visit an application system, the web server will send information about the user and store the information in the form of cookies on the client computer, which can be used to create dynamic and custom pages or store login information. If the web application system uses cookies, you must check whether the cookies work properly. The content of the test can include whether the cookies work or not and whether they are saved according to the scheduled time.

Through the above test means, the system has a good performance in network throughput, memory utilization and other aspects, and supports at least 2 500 concurrent access and concurrent operation, which has met the actual requirements of the system for the application of university environment. In terms of function, through the timely repair of the problems found in the test, through the function repair, all the functions designed at present can ensure normal use. Through the test of links, it is found that there are no dead links and invalid links in the system at present. Through the test of cookies, it is found that the use of this system will not affect the normal data saving functions of cookies under the windows series operating system environment and mainstream browser software environment. In short, the system can meet the needs of its work both in performance and function.

5. Conclusion

Based on the improved Apriori algorithm, this paper designs a dynamic early warning system for college students’ target course scores. In this paper, the overall development of the system, database design, functional design and other aspects of the detailed elaboration. The realization of key functions and technologies such as data manipulation and learning warning are described in the form of text and code explanation. Finally, the running effect and practical application performance of the system are described through the running example demonstration and system test. The system realizes four functional modules: process learning management, learning situation analysis, learning early warning and system management. It has a certain practical operation effect and has a certain practical application value.

References

Chartier

K.G.

Tiarsmith

E.C.

O’Shea

Kendler

K.S.

and Dick

D.M.

, College students’ use of strategies to hide facial flushing: a target for alcohol education, Journal of American College Health (1) (2019), 1–5.

Hafez

M.A.

Hanif

M.H.

Sidek

H.N.

Moustafa

Z.I.

Almkahal

and Sandra

J.A.

, Prediction of ultimate capacities of bored pile based on routine load tests and static analysis (case study), Advances in Industrial Engineering and Management 8(1) (2019), 7–9.

Xie

X.C.

Xue

Y.J.Y.

Zhao

Z.Q.

Chen

B.J.

and Jiang

, Risk prediction and factors risk analysis based on IFOA-GRNN and Apriori algorithms: application of artificial intelligence in accident prevention, Process Safety and Environmental Protection 122 (2018), 169–184.

Abraham

Dutta

Mandal

J.K.

Bhattacharya

and Dutta

, [Advances in intelligent systems and computing] emerging technologies in data mining and information security volume 813 (proceedings of iemis 2018, volume 2)

||

data mining in frequent pattern matching using improved Apriori algorithm, 10.1007/978-981-13-1498-8 (Chapter 61), 2019, 699–709.

Sahu

and Bisht

R.S.

, Improved Apriori algorithm for association rules using pattern matching, International Journal of Computer Ences and Engineering 7(7) (2019), 125–128.

Malakar

T.D.

Bhattacharyya

and Sarkar

S.K.

, Analytical surface potential based modeling of drain current & trans-conductance of work function engineered gate recessed S/D SOI/SON MOSFET, Advances in Industrial Engineering and Management 6(2) (2017), 107–110.

Xiao

Sun

and Li

, Research on fault-environment association rules of distribution network based on improved Apriori algorithm, IOP Conference Series Materials Ence and Engineering 569(3) (2019), 032079.

X.N.

Deng

F.Y.

Zou

Guo

and Fang

Y.C.

, Prediction of multiple diseases of soybean in complex environment based on improved Apriori algorithm, IOP Conference Series Materials Ence and Engineering 768 (2020), 052117.

Zhou

, Design and implementation of book recommendation management system based on improved Apriori algorithm, Intelligent Information Management 12(3) (2020), 75–87.

10.

Liu

Yang

Ren

Sun

G.Q.

and Li

Q.S.

, An improved Apriori algorithm based on matrix and double correlation profit constraint, in: Abstracts of the 4th International Conference of Pioneering Computer Scientists, Engineers and Educators (ICPCSEE 2018) Part I, 2018, p. 1.

11.

Roy

Bag

and Sarkar

S.K.

, CORDIC based robot navigation processor deploying RFID technology, Advances in Industrial Engineering and Management 6(2) (2017), 67–72.

12.

and Zhang

, Research on sensor network optimization based on improved Apriori algorithm, EURASIP Journal on Wireless Communications and Networking 2018(1) (2018), 258.

13.

Barot

P.A.

and Jethva

H.B.

, Unbalanced data classification using feature selection through bitapriori algorithm, International Journal of Computer Ences and Engineering 6(10) (2018), 701–704.

14.

Sornalakshmi

Balamurali

Venkatesulu

Krishnan

M.N.

Ramasamy

L.K.

Kadry

Manogaran

Hsu

C.H.

and Muthu

B.A.

, Hybrid method for mining rules based on enhanced Apriori algorithm with sequential minimal optimization in healthcare industry, Neural Computing and Applications 2020, 1–14.

15.

Sim

D.Y.Y.

The

C.S.

and Ismail

A.I.

, Improved boosted decision tree algorithms by adaptive Apriori and post-pruning for predicting obstructive sleep apnea, Journal of Computational and Theoretical Nanoscience 24(3) (2018), 1680–1684.

16.

Baoli

Ling

and Xiaoxue

, Knowledge based collection selection for distributed information retrieval, Information Processing & Management 54(1) (2018), 116–128.

17.

Ferreira

P.D.S.G.

Pegetti

A.L.

Piacesi

M.T.

Belderrain

M.C.N.

and Bergiante

N.C.R.

, Dynamic modeling of an early warning system for natural disasters, Systems Research and Behavioral Ence 37(2) (2020), 292–314.

18.

Nathan

H.L.

Duhig

Vousden

Lawley

Seed

P.T.

Sandall

Bellad

M.B.

Brown

A.C.

Chappell

L.C.

and Goudar

S.S.

, Evaluation of a novel device for the management of high blood pressure and shock in pregnancy in low-resource settings: study protocol for a stepped-wedge cluster-randomised controlled trial (cradle-3 trial), Trials 19(1) (2018), 206.

19.

Liu

X.L.

Jia

S.W.

and Wang

J.M.

, Comparison of user clustering based on user reputation and network structure, Computer Simulation 37(2) (2020), 221–225.