Research on the Application of Data Mining Technology in Educational Management System

Data mining technology has changed the way people understand data and analyze data to a large extent. The technology belongs to a new kind of information processing technology, which can provide information support to the decision makers and make for promotion of integrated development of educational reform. At present, the educational management system in colleges and universities mainly centers on the handling of daily routines, and the mass data merely realized the storage instead of exerting its function, besides, the significance hidden in the mass data often has not been discovered. Therefore, starting from the definition of data mining, the paper analyzed the necessity of application of data mining technology in the educational management system, and then studied the application of data mining technology in the educational management system.


Architecture of educational data warehouse system
The data warehouse is the collection of subject-oriented, integrated, stable data of different time, which is used to support the decision-making process in operating management. Before the construction of the data warehouse, it's a vital work to determine the architecture of the data warehouse system. Firstly, we should draw up a system framework of the data warehouse, which is suitable for the characteristics of the institution. The establishment of a truly practical, effective and flexible architecture of the data warehouse system is extremely essential, and the establishment of data warehouse system is to meet the increasing, more and more complicated statistics, analysis and forecast demands raised by managers at all levels. On the basis of maintaining the function of the existing application system, the system makes full use of the various data origins in the school, including faculty and staff data, teaching data, enrollment data, student management information, etc., carries out decision-oriented data reconstitution on them, and then transfers them into the data warehouse. With such data warehouse, the data mining application system and flexible inquiring system will be established facing the ultimate users based on decision support. Through this data warehouse system, the users can conduct deep statistics and analysis on teaching, curriculum provision and situation of student management from different perspectives, in order to direct their decision-making. The key technologies involved in the data warehouse are as follows: data extraction, data storage and management, data representation and data warehouse design. Generally speaking, the architecture of educational data warehouse mainly consists of two parts, i.e., the foreground and the background, which specifically includes data sources, data preprocessing, metadata, data mart, data warehouse, front-end application, etc.. The architecture of the educational data warehouse system is as shown in Figure 1.

Association rules mining in the educational administration system
Among the large amount of data mining research techniques, the most popular is the association rules mining. Association rules mainly refer to discovering the inner link between the objects and the basic idea is to determine the minimum support degree, find out all the frequent item sets (referred to as frequency set) larger than the minimum support degree, and then use the discovered frequency set to generate the desired rules. The core of association rules mining is recursion of the frequency set. The main algorithms of association rules mining contain time sequence discovery, sequence pattern discovery, association discovery and so on. 3.2.1 Association rules technology Data mining can be viewed as a way to discover knowledge in large database. As a new technology, data mining emerges following the development of database technology and it allows users to access or deal with a great deal of information. The purpose of data mining is to generate smart tools for evaluating and extracting information from the database. In recent years, data mining has caused great attentions of the education administrator and the administrative staff play an important role in higher education. Data mining involves a wide range of disciplines and fields and has many commonly used pattern discovery techniques, such as the decision-making tree and clustering technology, which belong to achievements of the field of machine learning and research. The emergence of association rules mining technology has greatly expanded the research of data mining and exerted huge impact on the field of data mining, thus the technology has become a vital branch of data mining research. Association rules can effectively discover the important association relations between data, besides, the expression form of the rules is relatively simple, which is easy to be understood and explained, resulting in the fact that association rules mining from large database has become a popularity in the field of data mining research in recent years.
Journal of Education and Practice www.iiste.org ISSN 2222-1735 (Paper) ISSN 2222-288X (Online) Vol.11, No.8, 2020 3.2.2 Apriori algorithm of association rules Apriori algorithm is one of algorithms of association rules. The first traversal of Apriori algorithm only calculates the number of specific index of each item to determine the large one item set. The subsequent traversal contains two stages. Firstly, candidate item set Ck is generated using the large item set Lk-1 discovered in the (k-1) traversal and the Apriori-gen function. Then the database is scanned to calculate the support degree of the candidate item set in CK, thus getting the order k large item set LK, whose support degree in Ck is no less than the minimum support degree. The above steps will be repeated until the large item set of some order becomes empty, then the algorithm stops. Description of Apriori algorithm is as follows： 1)1L ={large one item set}; 2) for( k =2; =Lk-1≠ɸ; k ++) do begin 3)Ck=apriori-gen(L k-1);//construct candidate item set 4) for all affairs t =D do begin 5)Ct =subset(Ck , t ); //search candidate item set contained in affair t 6) for candidate item sets t=Ct do begin 7) c.count++; // calculate support degree 8) end 9)L k={c =Ck|c.count ≥min sup}; //obtain order k large item set 10) end 11) answer=U k L K So far, there are a lot of algorithms of association rules, but all of the mining algorithms can be divided into the following steps regardless of the data structure, complexity or efficiency: firstly, carry out preprocessing on data related to the mining task. According to the specific requirements, carry out corresponding operation on the database, so as to form a normalized achievement database; secondly, for the achievement database, obtain all the item sets that meet the minimum support degree, i.e., the large item set. Since under normal circumstances, the database we are faced with is relatively large, this step is the core of the algorithm; thirdly, generate rules that meet the minimum confidence and form the rule set; fourthly, explain and output.
Nevertheless, the second step is the focus and difficulty of mining association rules, so most of the researches focused on this step.

Application of association rules in reasonable curriculum setting and learning evaluation
Association rules are used to describe the relations between data items in the database, namely, digging out the hidden association or relations between the data according to the appearance of some of the items of an affair. The students' curriculum study is progressive, and certain correlation and sequential order relation exist between the curriculum. We ought to know which courses should be taken before learning a more advanced course, because if the prerequisite course was not well studied, it's bound to affect the study of follow-up courses. The major setup and course system of colleges and universities should be synchronized with the social needs, which requires the schools to revise the teaching program in a timely manner and replace the curriculum unfit for the market requirements. In this way, we can take advantage of the data mining and adopt the correlation analysis, mainly the association rules to analyze the reasonability of curriculum setting. Carrying out evaluation on students is a link of the educational process, therefore, the function of evaluation is consistent with the educational objectives. Highlighting the developmental function of evaluation intensively reflects the educational idea of "all for the development of students". The students are undergoing a process of continuous development and changes, so the significance of education is to guide and promote the development and improvement of students. The development of students needs targets, guidance and incentives. Developmental evaluation determines individualized developmental goals for students. In nowadays, network of teaching and scientific research is universal, so we can use the data mining tools to analyze and process the achievement database and behavior record database to determine the strengths and weaknesses of the students and put forward specific, targeted suggestions on this basis. The topic of association rules is important in the field of data mining. It seeks the frequent patterns, correlation, dependency or causal structure between the item sets or between the object sets in the transaction data, relational data or other information carriers. Through the analysis of relations between data or records, it decides which things will happen together. It's one of the principal modes for current data mining research, it focuses on determining the relations between different fields of the data and finding out the dependencies between multiple domains that satisfy the given support degree and confidence level threshold. The basic model of association rules mining is shown in Figure 2： Journal of Education and Practice www.iiste.org ISSN 2222-1735 (Paper) ISSN 2222-288X (Online) Vol.11, No.8, 2020 Figure 2 Basic Model of Association Rules Mining By using different data mining methods in different modules of the educational management system, we can discover the relevant valuable information and knowledge. For example, in the topic of exam results analysis, we would like to analyze the incidence relations between the student's prerequisite course achievements, gender, assignments or classroom performance and the final test achievements, as well as the sequential relations between the courses in the teaching program. Using association rules mining in the student evaluation module can associate the test achievements of the students with the individual information (e.g.,gender, nationality, age, origin of student, etc..), thus applying the discovered rules into decision-making and providing valuable reference for qualityoriented education and teaching modes. Application example of the association rules is as follows: Assuming that I is a set of m courses,Database D is given as the set of achievement records of the students. Every Record T is a set of attributes in I. T has an exclusive identifier TID. Taking the achievement Record Table  1 of a group of students as an example, Table 2 is further analyzed by association rules.  A According to the above idea of carrying out improvement on the algorithm, the structure of candidate item set contains a list of transaction identifiers, and the length of the list is indeterminate, so the implementation is difficult in the relational database. Therefore, we did the following adjustments in the implementation of the algorithm: in the candidate item set table, each of the <item set, transaction identifier> is regarded as a record, in this way, one item set in the candidate item set can correspond to multiple records. Taking discovery of the association rules for student curriculum for example, part of the data fields of the course achievement table of students major in computer engineering is listed in Table 1, and some data of the table is listed in Table 2. However, the candidate item set obtained by processing data in Table 2 is listed in Table 3. Table 3 List of Candidate Item Set  Item  Value  ID  Item  Value  ID  GG01  A  001  GG03  B  003  GG01  A  002  GG03  A  004  GG01  B  003  ZY01  A  001  GG01  A  004  ZY01  A  002  GG02  B  001  ZY01  B  003  GG02  A  002  ZY01  B  004  GG02  C  003  ZY02  A  001  GG02  B  004  ZY02  B  002  GG03  A  001  ZY02  B  003  GG03  A  002  ZY02 A 004 Then we carry out statistics on the count of each item set through the candidate item set table and insert the count into frequent item set table. Meanwhile, the infrequent item sets in the candidate set are deleted to facilitate the formation of the candidate item set of the next level. Next, we use the educational management system-course relevant subsystem to generate frequent item set according to the achievement data. Thus it can be seen that the data warehouse in colleges and universities is a complex and huge project, which involves all kinds of institutions and personnel in the schools, thus needs plenty of investigations and data collections and sorting as well as a longer period for establishment and improvement. However, by building a data warehouse system, colleges and universities can give full play to the roles of all kinds of resources in the schools, so as to enhance the school management and decision-making level.

Conclusions
In a word, the data mining technology is a useful tool, which has been applied in multiple fields and obtained satisfactory results. In the field of higher education management, with the ever-increasing data information, the application of DM technology into the educational management system can promote the further reform, improvement and development of the education management. Of course, data mining itself is not almighty and its application can not be separated from the practical background. Only by following the people-oriented principle, will the technology have real vitality. Anyway, the educational management system contains a large amount of useful information to be discovered. In nowadays, Rejuvenate China through Science and Education is vigorously promoted, so data mining technology will definitely have important guiding significance for schools to develop better student training programs.