Evaluation Method of Effectiveness of College Students’ Employment Guidance Strategy Based on Data Mining

Aiming at the problem of low confidence in the effectiveness evaluation method of traditional college students’ employment guidance strategy, an effectiveness evaluation method of college students’ employment guidance strategy based on data mining is designed. The effectiveness evaluation index system of college students’ employment guidance strategy is constructed. The system is divided into three levels: target level, criterion level and index level, and the analytic hierarchy process is used to give weight to each index. According to the importance of the index, the effectiveness of college students’ employment guidance strategy is evaluated based on data mining. The experimental results show that the confidence of the designed evaluation method is significantly higher than that of the control group, which can solve the problem of low confidence of the effectiveness evaluation method of traditional college students’ employment guidance strategies.


Data Mining
With the continuous development of database and the wide application of system, like banks, insurance companies, securities companies and most financial service institutions, many organizations or enterprises have accumulated a large amount of data in their daily business activities (Justin & Ellen, 2019). However, most organizations or enterprises do not convert their data into valuable capital. The reason for this situation is that the information hidden behind these data is not so easy to be found (Pan & Center, 2019). In order to effectively participate in today's market competition, managers must find and reasonably use the valuable information hidden behind the historical data. The process of discovering and using the information hidden behind historical data is data mining.
Data mining refers to a non trivial process of extracting effective, implicit, potentially useful and ultimately understandable information and knowledge from a large amount of data in the database (Zhu, 2019). Data mining is a decision support process closely related to computer science. It can fully and automatically analyze the data of organizations or enterprises and make reasonable judgment and logical reasoning through machine learning, expert system, pattern recognition, database technology, mathematical statistics, parallel computing, visualization technology and other methods. And then dig out the potentially valuable information and knowledge to help management decision-makers adjust market competition strategies, reduce unnecessary risks and make favorable decisions. According to the difference between statistics and non statistics, data mining can be divided into two different types: hypothesis driven and discovery driven data mining (Zhou, Dan, Wen-Li, et al., 2019). The biggest difference between the two is the process of obtaining information.
Data mining is a complete and cyclic process of human-computer interaction. It is usually composed of multiple interrelated steps, such as definition problem, data selection, data preprocessing, data transformation, data mining, pattern interpretation, knowledge evaluation, etc. The schematic diagram of data mining process is shown in Figure 1.  As shown in Figure 1, the corresponding steps in the process are described as follows: (1) Defining problems: accurately finding out the definition of the problem to be mined and clarifying the theme and goal of mining are the cornerstone of the success of data mining. Before preparing useful data, data mining related personnel must clearly understand the target needs of end users, clarify the background knowledge, understand the relevant situation in relevant application fields and the data mining algorithms to be adopted in the process of data mining (Feng & University, 2019). Generally speaking, the final result of data mining is unpredictable, but the problems that need to be explored and mined can be known in advance. The so-called data mining for data mining has a certain blindness, and it is difficult to succeed after all (Xie, QI, Zhang, et al., 2019). Therefore, in the problem definition stage, the close cooperation between data mining personnel and relevant application experts and end users is particularly important.
(2) Data selection: the key of data selection is to determine the target data. According to the needs of end users and under the guidance of experts in relevant application fields, relevant data or samples are extracted from the original database to form a new meaningful target data set. In this process, some database operations are mainly used to process the original data set.
(3) Data preprocessing: further process the data determined in step (2), and check the consistency and integrity of the extracted data set. Remove the useless data deviated from the mining target, clean the noise data, and complete the data type conversion. According to the data characteristics and changes of the selected data set, the vacancy data are deduced and calculated by statistical methods.
(4) Data transformation: reprocess the preprocessed data according to the target task of data mining, so as to eliminate the redundant features in the data. The usual approach is to reduce the number of variables or decision attributes in the data set by using projection or other operations of the database.
(5) Data mining (a) Algorithm selection: further determine the type of knowledge to be mined through step (1), and then select the corresponding data mining algorithm. Because the data mining algorithms used for different knowledge types are different, such as association rules, classification, clustering, summary, etc., the expression forms of their results are completely different (Li, 2019). After determining the algorithm, selecting the appropriate data mining algorithm also includes selecting the appropriate model and parameters. There are two methods to select the algorithm: one is to select the corresponding algorithm according to the different data characteristics. The other is based on the needs of users. Some users need descriptive results, while others want to get results with as high prediction accuracy as possible.
(b) Establish the analysis model: use the previously selected algorithm to collect data from the preprocessed target data set. Extract the knowledge and information that users are interested in and show it to users in an intuitive and clear form (Lin & Raza, 2019). This is a very important link in the whole process of data mining.
(6) Pattern interpretation: explain the knowledge and information discovered in step (5). Remove redundant or irrelevant knowledge and information found by machine or user evaluation. In addition, if  Table 1, it is the evaluation index system of the effectiveness of college students' employment guidance strategy.

Calculate the Evaluation Index Weight of the Effectiveness of College Students' Employment Guidance Strategy
On the basis of constructing the evaluation index system of the effectiveness of college students' employment guidance strategies, calculate the weight of the evaluation index of the effectiveness of college students' employment guidance strategies. This paper adopts the method of expert scoring, in which the degree of compliance between professional setting and social needs, the signing rate at the end of each school year, the average student funds per academic year, the number of employment personnel per 100 students and the number of employment policy lectures per department student per academic year are selected as the key indicators of this evaluation, and their weights are calculated.
(1) Degree of fit between specialty setting and social needs. This index can be expressed by the ratio of the number of graduates of a certain specialty to the number of graduates required by the society, which reflects the adaptability of the specialty setting of colleges and universities to the social needs.
Calculation method: In formula (1), I refers to the degree of compliance between professional setting and social needs; A is the number of graduates of a major; B is the number of graduates required by the society.
(2) The signing rate at the end of each school year refers to the ratio of formal signing and flexible employment of colleges and departments by the end of each school year. The larger the index, the higher the employment level of colleges and departments. Calculation method: In formula (2) Thus, the weight of the evaluation index of the effectiveness of college students' employment guidance strategy is obtained.

Effectiveness of Employment Guidance Strategies for College Students Based on Data Mining
Based on the above calculated evaluation index weight of the effectiveness of college students' employment guidance strategy, judge the effectiveness of college students' employment guidance strategy. However, due to various reasons, the data in the original data set is often incomplete, which may contain some false information or wrong information. In addition, some data may be meaningless for data mining analysis, and so on. Therefore, before data mining, it is often necessary to do appropriate data preprocessing for the original data, otherwise unexpected error results may occur. In view of the characteristics that the data set attributes after mining, analysis and preprocessing of employment information data are all classification attributes, this experiment plans to use predictive classification algorithm to construct classification model for analysis. Classification is an important data mining method. Its purpose is to build a classification model (often called classifier) according to the characteristics of sample data. The model can map some new samples to a given category. The classification methods mainly include decision tree, k-nearest neighbor method, support vector machine, Bayesian method and neural network method. Because the decision tree is easy to understand and implement, its classification model is easy to test and evaluate, and has high reliability. In addition, the decision tree can mine feasible and effective patterns from large data sets in a short time. Therefore, the decision tree method is used to construct the classification model.
After the above analysis, this paper intends to use ID3 algorithm and improved ID3 algorithm to construct the corresponding decision tree classification model for the same data set. The results of each model are compared, and then a classification model with good classification effect is found for the data mining analysis of employment information data set, and the corresponding results are obtained, so as to provide decision support for college students' employment guidance.

Evaluation on the Effectiveness of College Students' Employment Guidance Strategy
This paper tests the consistency according to the effectiveness data of college students' employment guidance strategies. Set explicit variables that can be used for direct measurement to generate dynamic evaluation formula. Get the strong and weak relationship between each evaluation index and the effectiveness of the final college students' employment guidance strategy. The effectiveness evaluation formula of college students' employment guidance strategy based on data mining is shown in formula (6). www.scholink.org/ojs/index.php/jecs Journal of Education and Culture Studies Vol. 5, No. 4, 2021 Published by SCHOLINK INC.
Formula (6) is the evaluation result of the effectiveness of college students' employment guidance strategy based on data mining. Combined with the effectiveness evaluation system of college students' employment guidance strategies and corresponding index factors, each factor adopts the Crete 5-level positive scoring system in the survey. So far, the effectiveness evaluation of college students' employment guidance strategy has been realized.

Experimental Preparation
The research data of this paper comes from the employment information database of the employment guidance center of a university in Chongqing, mainly based on the employment information data of a certain session of graduates. The original dataset is an excel format file containing 4927 records. The attribute fields mainly include college, student number, graduation destination, company name, company affiliation, company address, company nature, employment status, remarks, company, dispatch certificate number, file processing method, gender, educational background, major, teacher logo, training method, school system, family, political outlook, enrollment time, graduation time, original company name, place of origin, candidate number, file receiving unit, QQ number and contact number. It takes the effectiveness of guidance strategy as the evaluation content by designing an example analysis. Firstly, it is used to evaluate the effectiveness of guidance strategies. The confidence is tested and recorded by Kinect software, and it is set as the experimental group. Then use the traditional methods to evaluate the effectiveness of college students' employment guidance strategies.
Similarly, test and evaluate the confidence through Kinect software, record it, and set it as the control group. It can be seen that the main content of this experiment is to test the evaluation confidence of the two evaluation methods. The higher the evaluation confidence value, the higher the evaluation accuracy of the evaluation method. Through 6 experiments, the experimental data were recorded according to the evaluation confidence measured in the experiment.

Analysis and Conclusion of Experimental Results
Compare the evaluation confidence under the two evaluation methods, and the comparison results of evaluation confidence are shown in Table 2   As shown in Table 2, the confidence of the evaluation method designed in this paper is significantly higher than that of the control group, which has obvious advantages.

Conclusion
There are still many meaningful topics about the application of data mining technology in college students' employment guidance and the research of decision tree classification algorithm. Next, the further work directions are as follows: Although the improved ID3 algorithm overcomes the disadvantage that the original ID3 algorithm prefers attributes with a large number of values to a certain extent. However, it is highly targeted and can not be widely applied to many fields. How to solve this problem and establish a more perfect decision tree classification model needs further research.
This paper only uses a single employment information data of college students, but does not involve some data similar to personal basic information and achievement information of relevant college students. In fact, these data have an impact on employment. Therefore, in the next step, building a more complete data warehouse, data mining and analyzing the impact of employment factors is also a topic worthy of research. For the same data set, different mining topics can be found, and the preprocessing methods and modeling algorithms will be different. Next, using other data mining algorithms such as association rules to mine the employment information data set is also worth studying.