Exploring the Integration Path of Student Management and Ideological and Political Education in Colleges and Universities

In the context of the new era, improving the quality of ideological and political education has become a key demand for the development of the socialist cause with Chinese characteristics in the new era. Under the comprehensive leadership of the Party, ideological and political education in colleges and universities has been deeply integrated with other fields and has realized a remarkable development leap [1]. Ideological and political education in colleges and universities and student management work support each other, together constituting the main battlefield and main channel of ideological education for college students [2]. Through the organic integration and linkage of the two, colleges and universities have formed a fruitful parenting synergy, promoting the overall development of students, and realizing the goal of cultivating talents for the party and nurturing the future for the country [3-4].

Despite the difference in content between ideological and political education and student management, they have a high degree of intrinsic fit in the goals and methods of educating people [5]. Ideological and political education focuses on shaping students’ ideological values, including socialist core values education and Marxist theory education [6-7]. Student management, on the other hand, covers the overall development of students, including enrollment, employment, school registration, academics, daily life, etc. [8-9]. The fundamental goal of the two is the same, i.e., to cultivate socialist builders and successors with all-round development of morality, intelligence, physicality, aesthetics and labor. This fit not only provides the possibility and inevitability for the combination of the two, but also provides a solid guarantee for colleges and universities to realize the goal of educating people [10-12]. Ideological and political education provides students with correct values and ideological guidance, while student management promotes the formation and development of students’ thoughts by regulating and guiding their behavior and creating a good learning environment [13-15]. Through the implementation of education and guidance, student management helps to promote the effective implementation and in-depth promotion of ideological and political education [16-17]. Therefore, the effective combination of ideological and political education and student management work has important practical and research significance.

This paper firstly introduces two algorithms of data mining technology as a theoretical method, on the one hand, by cleaning and analyzing the data such as the total amount of consumption, the number of times, and the average consumption value of the students’ cafeteria. Then it carries out data mining analysis using ID3 decision tree algorithm to find out the characteristic value of poor students and get the prediction model of poor students. And borrow the poor students assessment results of previous years to assess whether the rules formed by data mining are effective. And on this basis, the strategy of combining student financial aid management and ideological and political education in colleges and universities is proposed. On the other hand, the indicator system of students’ personal information is constructed, and data mining technology is utilized to detect the correlation between each indicator and students’ performance as well as the change of students’ performance. Then based on the indicators of students’ feature library, respectively based on achievement and based on behavioral data, use the simple Bayesian classification algorithm to predict and analyze the students’ academic early warning. And on this basis, the functional realization strategy of ideological and political education in college students’ academic early warning is proposed, which promotes educational administrators and teachers to work better and helps students to complete their studies better.

2

Data mining techniques

2.1

ID3 algorithm based on information theory

2.1.1

Application of information theory to the ID3 algorithm

So far there have been many kinds of decision tree generation algorithms, but the most influential international example learning algorithm is the first ID3 (Iterative Diehotomieversion3) algorithm.

First, a numerical metric is applied to the information: 1 $I = \log_{2} S^{n} = n \log_{2} S$

The important concepts of information theory are:

1)

Self-information

Suppose X₁,X₂⋯X_n is a signal from a source. Prior to receiving X_i, the recipient’s uncertainty about the outgoing X_i is defined as the amount of word information in the message symbol I(X_i), I(X_i) = –logP(X_i).

where P(X_i) is the probability that the source sends. If 2 is used as the base, the resulting information quantity is in bits.

2)

Information entropy

While the self-information quantity reflects the uncertainty of the symbol, the information entropy can be used to measure the uncertainty of the whole source. The definition is as follows: 2 $\begin{matrix} H (X) = P (X_{1}) I (X_{1}) + P (X_{2}) I (X_{2}) + \dots + P (X_{n}) I (X_{n}) \\ = - \sum_{i = 1}^{n} P (X i) - \log P (X i) \end{matrix}$

where n is the number of all possible symbols for the information source X, i.e., the information entropy (average information) is defined in terms of the average amount of self-information provided by the source per symbol sent.

3) Conditional entropy. If source X and random variable Y are not independent of each other and the recipient receives message Y, then conditional entropy H(X/Y) is used to measure the uncertainty that still exists about random variable X after the recipient receives random variable Y. Let X corresponds to the source symbol X_i, Y corresponds to the source symbol Y_j, P(X_i/Y_j) is the probability when Y is Y_j and X is X_i, then we have: 3 $H (X / Y) = - \sum_{i = 1}^{n} \sum_{j = 1}^{m} P (X i / Y j) \log P (X i / Y j)$ where i∈[1,n], j∈[1,m]

The conditional entropy is less than the information entropy, which means that the average uncertainty about the input symbols X is reduced after receiving all the symbols of the symbol set Y. That is to say that some uncertainty about the sender is always eliminated after the message has been transmitted, and thus some information is gained.

4) Average mutual information. It is used to indicate the amount of information about X that signal Y can provide, and is denoted by I(X/Y): 4 $I (X / Y) = H (X) - H (X / Y)$

Following the definition of information theory, it is useful to assume that the set of training samples is X, which contains IXI data samples, and that the category attribute can take on n different values corresponding to n different categories C_i(i = 1,2,⋯,n). P(C_i) is the probability that any data object belongs to a category C_i, which can be computed as C_i/XI. Then the amount of information required to categorize a data object is: 5 $H (X / Y) = - \sum_{i = 1}^{n} P (C_{1}) \cdot \log P (C_{i})$

The decision tree learning process is the gradual reduction of uncertainty about the division using a decision tree. If the test attribute A is selected for testing and A takes v different values {a₁,a₂,⋯,a_v}. Then the conditional entropy of dividing the current sample set using a specific take a_j of attribute A is: 6 $\begin{matrix} H (X / A) = - \sum_{j = 1}^{n} \sum_{j = 1}^{v} P (C_{1}, A = a_{j}) \cdot \log P (C_{1} / A = a_{j}) \\ = - \sum_{i = 1}^{n} \sum_{j = 1}^{v} P (A = a_{j}) \cdot P (C_{1} / A = a_{j}) \cdot \log P (C_{1} / A = a_{j}) \\ = - \sum_{j}^{v} [P (A = a_{j}) \cdot \sum_{i}^{n} P (C_{1} / A = a_{j}) \cdot \log P (C_{1} / A = a_{j})] \end{matrix}$

From the definition of mutual information in Eq. (4), it can be calculated that the amount of information I(X,A) provided by attribute A for categorization is: 7 $I (X, A) = H (X) - H (X / Y)$

The smaller the value of Eq. (6), the larger the value of Eq. (7), indicating that the greater the information provided by the selection of test attribute A for classification, the smaller the degree of uncertainty about classification after selecting A as the test attribute.The ID3 algorithm is to select the attribute that makes I(X,A) the largest attribute as the test attribute and call I(X,A) the amount of information provided by test attribute A for classification, i.e., the information gain brought by attribute A.

2.1.2

Principles of the ID3 algorithm

The D3 algorithm is a typical decision tree learning algorithm that centers on the use of the information gain method as an attribute selection criterion at all levels of the decision tree node, which is used to determine the appropriate attribute to be used in generating each node. This allows the attribute with the highest information gain to be selected as the test attribute for the current node so that the minimum information is required to classify the subset of training samples obtained by dividing using that attribute.

The basic principle of ID3 is as follows: let E = F1 × F2 ×⋯×Fn be a n-dimensional exhaustive vector space, where F_j is a set of exhaustive discrete symbols, and an element $e = 〈 v_{1}, v_{2}, \dots, v_{n} 〉$ in E, called an example, where v_j ∈ F_j,j = 1,2,…,n. Let PE and NE be two sets of examples of E, called the positive example set and the negative example set, respectively. Assuming that the size of the positive example set PE and the negative example set NE in the vector space E are p and n, respectively, ID3 is based on the following two assumptions:

1) The classification probability of a correct decision tree on the vector space E for any sample set is the same as the probability of positive and negative examples in E.

2) According to Eqs. (2)-(5), the amount of information required for a decision tree to be able to make a correct category judgment on a sample set is: 8 $I (p, n) = (\frac{p}{p + n} \log_{2} \frac{p}{p + n} + \frac{n}{p + n} \log_{2} \frac{n}{p + n})$

3) Information entropy. The best assessment of impurity is the average information entropy, if attribute A is chosen as the root of the decision tree and A takes V different values {A₁,A₂,⋯A_v}, using attribute A can be divided E into v subsets {E₁,E₂,⋯E_v}, where E_i contains sample data in E which attribute A takes A_i values. Assuming that E_i contains p_i positive and n_i negative examples, the desired information required for subset E_i is I(p_i,n_i) and the information entropy required with attribute A as the root is: 9 $E (A) = \sum_{i}^{v} \frac{p_{i} + n_{i}}{p + n} I (p_{i}, n_{i})$ where $I (p_{i}, n_{i}) = - \frac{p_{i}}{p_{i} + n_{i}} \log_{2} \frac{p_{i}}{p_{i} + n_{i}} - \frac{n_{i}}{p_{i} + n_{i}} \log_{2} \frac{n_{i}}{p_{i} + n_{i}}$

4) Information Gain. The degree of information gain of an attribute is the value of the period of entropy abatement after partitioning by that attribute, with A as the root of the information gain that is: 10 $G a i n (A) = I (p, n) - E (A)$ ID3 selects the attribute with the largest attribute Gain(A), which is the smallest attribute E(A), as the root node.

2.2

Plain Bayesian classification algorithm

Simple Bayes is an algorithm developed on the basis of Bayesian theory, it is mainly based on Bayes’ theorem and independence theorem, it is very good to understand and master, simple and easy to understand at the same time the classification effect is very good.

Many of the classification algorithms in machine learning are discriminative algorithms, which analyze and study the outputs and inputs by finding their decision-making relationships. And the plain Bayesian algorithm looks for the joint distribution P(X,Y) of output Y and variable X and then uses P(Y\X) = P(X,Y)/P(X) to arrive at the final classification result.

From the knowledge of probability theory we know that the formula for conditional probability is:

P(XY) = P(Y\X)P(X) = P(X\Y)P(Y), which states that the probability of event X and event Y occurring simultaneously is equal to the probability of Y occurring conditional on X occurring, or the probability of X occurring conditional on Y occurring.

Bayes’ theorem is to know the probability of a certain condition, get the probability of the event after the swap, that is, in the case of known P(X|Y) to get P(Y|X). Where P(Y) is called the base probability (also called a priori probability), P(Y|X) is called the a posteriori probability of Y, that is, after the known X can be judged after the probability of Y occurs. In real life, we can often get P(X|Y) directly or indirectly, but it is difficult to get P(Y|X), so we must use P(X|Y) as a bridge to get P(Y|X), and Bayes’ theorem can be formula (11) for: 11 $P (Y | X) = \frac{P (X | Y) P (Y)}{P (X)}$

The results obtained by the Plain Bayesian method are presented in the form of probabilities, so the samples to be classified in the test set are classified into different types according to the magnitude of the probabilities they have obtained.

Plain Bayesian classification generally has the following steps:

1) Data preparation. This stage is mainly to prepare for the subsequent classification, including the acquisition of the dataset, the parsing and delineation of the sample feature attributes, all the input data to be classified, the output features and the training samples.

2) Training the classifier. This stage is based on the frequency of occurrence and attribute features of each class of samples in the previous stage to divide and train to get the classifier.

3) Application testing. This stage classifies the target problem according to the model obtained from the algorithm and obtains the classification results of the target features.

Suppose X is a probability space consisting of mutually independent events, X = {b₁,b₂,…b_n} and the objective function ∫(x takes values from a finite set V with sets V representing different categorization categories. After the plain Bayesian model is trained through the training set, it predicts the maximum probability category V_MAP in the new sample for the newly given sample attribute values {b₁,b₂,…b_n}.

12

V_{M A P} = \arg \max_{v_{i}, V} P (v_{j} b_{1}, b_{2} \dots b_{n})

Based on Eq. (12) above, Bayes’ theorem can be rewritten as: 13 $\begin{matrix} V_{M A P} = \underset{v_{1} \in V}{\arg \max} \frac{P (b_{1} b_{2} \dots b_{n} | v_{j}) P (v_{j})}{P (b_{1}, b_{2} \dots b_{n})} \\ = \arg \max_{v_{i} \in V} P (b_{1}, b_{2} \dots b_{n} | v_{j}) P (v_{j}) \end{matrix}$

According to Equation (13), P(v_j) can be obtained, which is the frequency of occurrence of the data to be categorized in the sample. However, estimation P(b₁,b₂…b_n|v_j) is not feasible. This is because the individual features in the probability space X are independent of each other when the target value is known. Therefore, we can convert Eq. (13) into the following formula, where P(v_j) represents the frequency of occurrence of the target category for classification, $\prod_{i}^{n} P (b_{j} | v_{j})$ represents the product of the conditional probabilities of occurrence of each attribute under a particular target category, and V_NB represents the final result of the algorithm.

14

V_{N B} = \arg \max_{v_{i} \in V} P (v_{j}) \prod_{i}^{n} P (b_{j} | v_{j})

The main advantage of simple Bayes is that the algorithm has good classification efficiency; the model effect is better when the amount of data is small, and it can deal with multi-classification tasks; it is not accurate enough to deal with the missing values, and the model is relatively simple, commonly used in text classification. However, there are some disadvantages, if the data volume is large or the data correlation is strong, the model effect is limited.

3

Integration and Application of Financial Aid Management and Ideological and Political Education in Colleges and Universities

3.1

Application of Data Mining Techniques in Smart Funding

3.1.1

Data pre-processing

1) Establishment of relevant data standards: according to the industry information regulations required by the Provincial Department of Education, the actual data and information situation of the university is integrated to establish data standards that meet the practical application.

2) Irrelevant data removal: including business system auxiliary fields and non-undergraduate student data that are not related to behavioral records.

3) Duplicate data removal: generic de-duplication processing is realized by kettle.

4) Normative checking: including enumeration type and regular type, normative data are directly entered into the database, and irregular data are supplemented with manual review (e.g., ethnicity).

5) Consistency checking: mainly for overlapping or associated fields of different data sources, the highest priority of academic affairs data.

Targeted cleaning: many data issues are closely related to the business use of its related systems and its own data characteristics, resulting in relatively rough data. Part of the data in the system there is duplication, incompleteness and other data defects it is necessary to target cleaning. Including de-duplication operations and delete invalid data, eliminate data redundancy, eliminate abnormal data and incomplete data can not be used to ensure data validity. The purpose of targeted cleaning is sufficient to some of the key missing information to make up for the times, and to help the extraction of time-series features completed in advance - part of the data preparation and conversion. School card system data in the data part of the data occurred many times repeated, missing data and other data defects, the need for data cleaning. To delete the duplicate data and remove the useless data, remove the redundancy of the data, eliminate the abnormal data and unusable data with missing data, in order to ensure the validity of the data. The cleaned data is shown in Table 1.

Table 1.

The data table after cleaning

Student cardnumber	Student grant	Total consumption	Maximum single consumption	Median consumption	Total canteen consumption	Average value of a single consumption	Total supermarket consumption	Consumption times	Canteen consumption times	Engel coefficient
341908	0	6293.3	200	1.7	1363.6	4.31	129.4	1228	167	0.217
342801	1500	4826.1	180	1.5	838.3	3.38	52.7	1072	162	0.174
343615	0	7510.4	200	2.1	1882.7	6.58	205.0	1454	193	0.251
343841	0	9381.8	320	3.4	2340.2	7.84	567.8	1572	269	0.249
345067	1500	5082.9	100	1.5	390.6	3.61	40.6	1002	58	0.077
346392	0	10372.5	250	3.5	3982.5	8.66	842.4	2051	183	0.384
347540	0	11842.4	300	4.3	3730.0	7.04	977.5	1905	152	0.315
348375	1500	6392.7	200	1.6	882.1	6.72	103.2	927	126	0.138

3.1.2

Analysis of experimental results

The construction of ID3 decision tree is the key to model the data, and any decision tree grows without limitation without pruning. And the number of decision trees will affect the effect of ID3 algorithm model’s widespread use to a certain extent. Also the number of decision trees will directly change the computational speed and classification results of ID3 algorithm, so the number of decision trees is crucial for modeling. If the number of decision trees is high, it may reduce the speed of ID3 algorithm: on the contrary, if the number of decision trees is low, it may reduce the classification accuracy of the cast model. It follows that a suitable number of decision trees needs to be captured. This is used to ensure that the relationship between the number of decision trees and the accuracy of the test set is established without varying the remaining parameters as shown in Figure 1.

Figure 1 shows that if the number of decision trees is in the middle of 200-300, the model accuracy is high, up to 70% or more. The optimal number of parameters of the decision tree in the algorithm of this thesis is 250, which ensures the accuracy of the modeling.

Adaboost method can improve the learning accuracy, it collects a number of weak classifiers and then vote to decide the classification, where each weak classifier has a different weight value, when encountering samples misclassified by previous weak classifiers, it will improve the weight value in the latter weak classifier, which will lead to the maximum possibility of being correctly classified. In this paper, the results obtained by using ID3 algorithm and Adaboost method classification will be compared in Table 2, the bold mark is the data of prediction error, it can be seen that in the test set, the prediction accuracy of the poverty level obtained by using ID3 algorithm reaches about 90% or so, and the effect of Adaboost classification reaches about 70%.The ID3 algorithm is compared to the Adaboost method classification Accuracy is higher, the determination of the results on the show better, the establishment of the model of the student’s grant prediction and the previous implementation of the assessment results of the grade is basically the same, indicating that the evaluation of the model is effective. Of course, because the quantity and quality of various data are not high, the dimension of mining is not high, so the accuracy of the conclusions obtained needs to be further verified.

Table 2.

Category prediction results

Student card number	Grants for students	ID3 prediction	Adaboost prediction
390372	2000	2000	2000
392384	0	0	0
393451	2000	2000	1500
393693	1500	1500	1500
394808	1000	1500	0
395265	2000	2000	2000
396781	1000	1000	2000
396974	0	0	0
397080	1500	1500	1500
397157	0	0	0

3.2

Integration Strategies of Student Financial Aid Management and Civic Education

In this study, data mining technology is used to collect, analyze, and mine all kinds of data information of students occurring in the school, and quantitatively assess the degree of poverty. The above section verifies the feasibility of this technology in smart financial aid in colleges and universities. Next, we continue to explore the integration path of combining financial aid management and ideological education in colleges and universities.

The primary service object of student financial aid work is students with family economic difficulties, which is a relatively special group. Affected by family economic factors, their hearts may be more sensitive, more fragile, more glassy than ordinary people, and students in colleges and universities are in the transition period from adolescence to adulthood, the physical and mental development of the transition period, which requires us to adhere to the basic spirit of the financial aid work in colleges and universities must be “student-oriented”, these students with financial difficulties in family This requires that the financial aid work of universities must adhere to the basic spirit of “student-oriented” and carry out ideological and political education for these students from economically disadvantaged families in a gentle and orderly manner, so as to help them become adults. However, in the actual process of work, the student financial aid work in colleges and universities is often only purely material and economic support, but neglected to educate their work.

1)

Strengthen the comprehensive awareness education, assisting the overall development of students

Colleges and universities in the student financial aid work at the same time, you can carry out integrity education, gratitude education, inspirational education and other thematic ideological education activities, through the relevant theme of the seminar, policy propaganda, thematic class meetings, excellent students inspirational cases of success in a variety of forms, multiple dimensions into the student’s daily learning activities, in order to improve the students’ sense of honesty, sense of gratitude and self-reliance and self-improvement consciousness.

Mobilize students to actively participate in various school activities, participate in all kinds of volunteer services and social welfare activities, improve their sense of existence, sense of belonging and sense of achievement, enhance students’ interpersonal skills and social adaptability, improve their ideological awareness and comprehensive ability, and assist students in comprehensive development.

2)

Improve the support system and cultivate a sound personality

It is especially important to understand the mental health of poor students. Schools can detect the psychological state of students, especially students from economically disadvantaged families, through regular mental health assessment and provide targeted mental health support and intervention. If students develop psychological problems, they should be guided to receive professional psychological counseling in a timely manner to ensure the stability of their mental health.

In addition, schools can provide poor students with relevant internship opportunities, provide internship units by creating additional positions such as work-study, and recommend suitable career plans for them, in order to help them gain practical experience and financial support outside of their studies.

3)

Establishing a sound ideological and political education mechanism in financial aid work

First of all, schools need to increase the staffing of full-time teachers in financial aid to ensure that they have enough time and resources to concentrate on financial aid work and closely integrate it with ideological and political education. These teachers will serve as guides for ideological and political education, and will be responsible for establishing contact with students and providing support and guidance.

Second, schools should raise awareness of ideological and political education in the financial aid work of students in higher education. This involves building a consensus that financial aid is part of the work of educating people and aims to improve the overall quality of students rather than just providing financial support. Therefore, schools need to emphasize the mutual collaboration between the financial aid management department and the school’s ideological and political education department to ensure that financial aid work is truly part of the work of educating people.

Finally, schools need to actively promote home-school cooperation in order to facilitate the co-education of students and families in ideological and political education. Schools can set up a family interview mechanism to regularly learn about the situation of poor students’ families and have positive exchanges with parents. In addition, schools can carry out family education support programs to provide families with relevant educational resources and assistance, and improve the educational level of families so as to promote the overall development of students.

In conclusion, the establishment of a sound ideological and political education mechanism in financial aid work requires the joint efforts of all school personnel and the close collaboration between the departments of financial aid work and ideological and political education.

4

Integration and application of students’ academic early warning and ideological and political education

4.1

Academic Early Warning Based on Student Behavior Analysis

4.1.1

Analysis of correlation between single indicators and achievements

To analyze the overall consumption data of students, the consumption data extracted from the card can be divided into two categories: cafeteria consumption and supermarket consumption. From the cafeteria consumption data, we can observe the regularity of students’ daily life and grasp the basic life and consumption of students. From the supermarket consumption data, we can judge students’ consumption habits and daily life expenditure ratio. In the first semester of the sophomore year, the total consumption data of the cafeteria and the total consumption data of the supermarket were used to correlate with the average credit performance of the students. Figure 2 and Figure 3 show the correlation analysis between cafeteria consumption and supermarket consumption and the average credit performance respectively.

Through the analysis, we get the correlation coefficient r between cafeteria consumption and average credit performance is 0.27, and the p-value of the significance test is 0.068, which indicates that the two have a certain degree of correlation; the correlation coefficient between supermarket consumption and average credit performance is 0.074, and the p-value of the significance test is 0.21>0.05, so it is considered that the two are basically not correlated.

Behavioral data to a certain extent reflect the students’ learning status, learning status can affect the learning changes, so consider the single indicator and the change of students’ performance for correlation analysis. According to the screened single indicators that are significantly related to grades, their correlation with changes in grades is analyzed. The comparison of correlation coefficients between single indicators and GPA and GPA change rate is shown in Table 3.

Table 3.

Correlation coefficient between single index and GPA and GPA change rate

Index	GPA	GPA change rate
Sex	0.25	-0.11
Early rise index	0.32	0.35
Canteen consumption	0.27	0.39
Days in the library	0.28	0.45
Library study hours	0.29	0.43
Breakfast punctuality rate	0.28	0.39
Lunch punctuality rate	0.43	0.35
Dinner punctuality rate	0.27	0.33

The data in Table 3 show that the rate of change in mean GPA for gender is -0.11 compared to its mean GPA of 0.25, and the rate of change in mean GPA for lunch positivity is 0.35, which is less than its mean GPA of 0.43. Except for these two indicators, the correlation coefficients of the remaining indicators are greater than the correlation coefficients with the rate of change in mean GPA, and it is reasonable to believe that the behavioral data indicators are more closely correlated with the changes in students’ grades are more closely correlated, providing some ideas for subsequent analysis.

4.1.2

Academic Early Warning Projections

Based on the plain Bayesian classification algorithm, the prediction of the change in pre-demon rank from the first three months of the second semester of the sophomore year to the second semester of the sophomore year was based on the first three months of the sophomore year’s behavioral data. Table 4 shows the results of predicting the change in pre-warning level.

Table 4.

Forecast results of the change in the forecast level

Predisplacement equalization index	Accuracy rate	Recall rate
2	0.37	0.31
1	0.70	0.92
0	0.99	0.94
-1	0.76	0.90
-2	0.28	0.24

The overall correct rate of predicting warning level changes based on the plain Bayesian classification algorithm is 90.41%, which has a high accuracy. In the prediction results of predicting the change of warning level based on the behavioral data in the first three months of the semester, the prediction effect is poor for those whose warning level change is -2 and 2, and some of the -2 is predicted to be -1 while 2 is predicted to be 1. For the group of smaller change of warning level (students whose indicators of change of warning level take the values of -1 and 1), the prediction results of the recall rate are all over 90%, with a recall rate is high but the precision rate is relatively poor.

The overall accuracy of predicting warning level changes based on behavioral data in the first three months of the semester is high, and the combination of predicting warning level changes on the basis of historical warning levels can be used to predict the next stage of warning levels in advance. In the classification prediction based on the logistic regression algorithm, the prediction effect on people with warning level changes of -2 and 2 is poor, which is obviously not in line with our expected goal, so we consider comparing the classification effect based on different machine learning algorithms, so as to improve the prediction effect on each category of people.

4.2

Integration Strategy of Students’ Academic Early Warning Management and Civic and Political Education

The previous paper used the plain Bayesian classification algorithm to carry out academic early warning prediction based on students’ behavior analysis. Next, this study discusses the strategy for realizing the function of ideological and political education in the academic early warning mechanism of college students.

The realization of the function of ideological and political education in college students’ academic early warning mechanism needs to rely on the improvement of the quality of ideological and political educators, the innovation of the synergistic elements of ideological and political education, and the orderly articulation of the various links of early warning.

1)

Improve the quality of ideological and political educators

First, sufficient patience and democratic working style. It is necessary to cultivate the confidence and determination of ideological and political educators to persevere and move forward. Educators need to be good at listening, listening to understand the students’ learning situation, living conditions, personality characteristics, ideological behavior, and then summarize the problem.

Secondly, deep professional knowledge and broad knowledge. Educators need to know more about the basic knowledge of the subject of academic early warning and the cutting-edge dynamics of the discipline, in order to lay a solid foundation for good academic early warning work.

Third, strong research ability and expression ability. To target and effectively do the ideological and political work of the academic early warning object, understand the subjective and objective reasons for the formation of these characteristics, and determine the development trend, so as to do the academic early warning work targeted.

2)

Innovative synergistic elements of ideological and political education

Promoting the function of ideological and political education in academic early warning requires the creation of a favorable atmosphere, the promotion of positive interaction between the ideological and political courses and courses of ideological and political education, and the coordination of school management and college students’ initiative. The synergistic effect of ideological and political courses and curriculum ideology and politics spreads ideas and truths for college students while spreading theoretical knowledge, shapes students’ character, morality, conduct and taste, and enables college students to establish the values that society needs.

Coordinate the school management with the realization of college students’ initiative. The following ways can be used to promote the realization of students’ subjective initiative: first, to enhance self-confidence with specialties or strengths. Counselors need to be good at discovering students’ strengths, creating a platform for displaying them, building up the self-confidence of students in difficulty, and guiding them to solve their learning difficulties so as to achieve all-round development. Second, the use of incentive mechanisms. In the academic early warning mechanism for college students, appropriate material and spiritual rewards are given to students in difficulty who have made significant progress and their supporting partners to recognize their hard work and efforts.

3)

Promote the orderly convergence of various aspects of ideological and political education

First, do a good job of humanistic care before the occurrence of academic warning. Teachers in the classroom need to improve the quality of classroom teaching, enhance the attractiveness of the classroom, give more care and encouragement to students with learning difficulties, answer more questions and solve their problems, and improve their confidence and ability in the subject.

Secondly, build an all-round and multi-angle ideological and political education synergy. After the fact of academic warning, counselors further targeted to understand the student thought dynamics, psychological condition, family emotional atmosphere and other information. Involving academic style problems to further in-depth interviews with the students themselves, involving psychological problems is feedback to professional psychological counseling teachers, teaching problems on the feedback to the relevant teaching leadership, family problems with parents to get in touch.

Third, the development of a rigorous support program to enhance academic interest. Student party members and outstanding student cadres set up support groups and interest activity groups to provide targeted academic support for students with academic difficulties.

Fourth, establish a feedback mechanism for timely communication. Throughout the warning period, counselors or class teachers to do timely communication with the target of early warning, make full use of modern information means and parents to establish regular dynamic contact, according to the performance of early warning to determine whether to revoke the warning or re-adjustment of the academic warning program.

5

Conclusion

In this study, the integration of student management and ideological and political education in colleges and universities is explored by utilizing two data mining techniques.

An experimental analysis of the ID3 decision tree algorithm in grant management shows that the number of decision trees is in the middle of 200-300, and the model accuracy is high, up to 70% or more.The number of optimal parameters of the decision tree in this paper’s algorithm is 250, which ensures the accuracy of the modeling.The prediction accuracy of poverty level obtained by using ID3 algorithm reaches about 90% or so, while the effect of Adaboost classification reaches about 70%.ID3 algorithm is compared to Adaboost method classification accuracy is higher, and the identified results show better. Through the application of ID3 data mining technology in financial aid management, it can better provide technical support for the integration of financial aid management and ideological and political education of students in colleges and universities.

The experimental analysis about the plain Bayesian classification algorithm in academic early warning prediction shows that the feasibility of using behavioral data to predict grade changes is verified through the correlation analysis of students’ behavioral data and test scores. Further, the overall correct rate of the proposed Parsimonious Bayesian classification algorithm based on Parsimonious Bayesian classification algorithm for predicting the change of warning grade is 90.41%, which has a high accuracy. The prediction system under this algorithm can better help the integration and application of academic early warning management and ideological and political education for college students.

Lingua:: Inglese

Frequenza di pubblicazione:: 1 volte all'anno
Argomenti della rivista:: Scienze biologiche, Scienze della vita, altro, Matematica, Matematica applicata, Matematica generale, Fisica, Fisica, altro

Feed RSS della rivista

Exploring the Integration Path of Student Management and Ideological and Political Education in Colleges and Universities

Ke Wang

Pubblicato online: 19 mar 2025

Ricevuto: 18 ott 2024

Accettato: 14 feb 2025

DOI: https://doi.org/10.2478/amns-2025-0521

Parole chiaveID3 decision tree, Plain bayesian classification algorithm, Financial aid management, Academic warning, Ideological and political education

© 2025 Ke Wang, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

Parole chiave
ID3 decision tree, Plain bayesian classification algorithm, Financial aid management, Academic warning, Ideological and political education