Student achievement prediction and auxiliary improvement method based on fuzzy decision support system

Abstract

This paper proposes a student achievement prediction model based on fuzzy decision support system (FDSS), which uses multi-dimensional data to accurately predict student achievement. A prediction model including students’ historical achievement, learning behavior and class participation is constructed, and fuzzy reasoning mechanism is used to deal with the uncertainty and fuzziness of data. Experimental results show that the FDSS model has high prediction accuracy on multiple data sets, and the error range is controlled at a reasonable level. Compared with traditional machine learning methods, FDSS model has advantages in prediction accuracy and generalization ability. This paper also discusses the application effect of the model in different subjects, grades and regions of the student data, and puts forward the model optimization and improvement strategies. Fuzzy decision support system provides an effective solution for student achievement prediction in the field of education, and provides scientific decision support and personalized guidance for education decision-makers and teachers.

Full text

Translate

Turn on search term navigation

Introduction

The fuzzy reasoning formula is the core mechanism of FDSS. It allows the model to handle uncertainty and vagueness in the input data. The prediction Y is determined by aggregating fuzzy rules. Each rule contributes to the final output based on its relevance, as determined by the fuzzy membership function and rule weights. In the field of student achievement prediction, the main focus is on using different technical means and models to improve the accuracy and efficiency of prediction.

Salamis proposed an optimized integrated deep neural network model, which combines multiple learning strategies and supports the implementation of personalized learning methods by accurately predicting student achievement [1]. In the application of decision support system, Govindan and Li proposed a decision support system based on fuzzy logic to automatically assess ergonomic risks [2]. In addition to deep learning models, Xiao et al. and Lee et al. focused on the prediction of online learning performance based on psychological and emotional factors. They used a BP neural network model to analyze the emotional fluctuations shown by students during the learning process to predict grades [3, 4]. Regarding the application of fuzzy decision support system in academic achievement prediction, Hoyos et al. demonstrated the application of fuzzy cognitive map in the medical field by studying clinical decision support system [5]. Burda et al. proposed a decision support system based on fuzzy cognitive maps and applied it to the decision support of community workers. Fuzzy modeling of complex decision problems helps decision makers make effective judgments in the face of uncertainty [6]. Nabizadeh et al. conducted a study on the early prediction of student achievement for gamified courses. Their research adopts a predictive model based on machine learning to identify the academic difficulties students may face in advance and provide data support for teaching adjustment [7].

This study builds a student achievement prediction model based on fuzzy decision support system, explores how to improve the accuracy of student achievement prediction through this model, and provides targeted improvement schemes. Fuzzy decision support System (FDSS) is adopted as the main technical framework, combined with data mining and machine learning methods, and multidimensional data analysis is used to build a student achievement prediction model [8]. Promote the application of fuzzy decision support system in the field of education, break the limitations of traditional performance prediction methods, and provide more scientific and efficient solutions. Fuzzy logic is introduced to more accurately reflect the formation mechanism of student achievement, identify the multi-dimensional factors that affect the achievement, and provide data support and decision-making basis for educators. The predictive model can provide effective means of student management for educational institutions and schools, help to formulate more targeted teaching strategies, and realize personalized education and precise teaching.

The methodology used in this study revolves around combining fuzzy logic and data mining techniques to predict student achievement. The system uses multi-dimensional data sources, including academic performance, learning behavior, family background, and other personal factors. The data is first pre-processed through cleaning, standardization, and feature selection. The feature selection process helps eliminate redundant variables and ensures that the most relevant ones are used in the prediction model. The fuzzy decision support system (FDSS) is then applied to process the data. The model utilizes fuzzy logic to handle the inherent uncertainty and complexity in the data, converting input data into fuzzy values and using a rule-based inference engine to predict the student’s achievement.

Materials and methods

Data source

Data sources mainly include two parts. The first part is the performance data of the students, such as the final exam results of each subject, the usual homework results, and the class performance. It reflects the performance of students in different learning activities and provides the basis for the achievement prediction. The second part is the data of other factors related to student achievement, such as students’ learning time, family background, class participation, mental health status, learning resource use, etc. These factors have a potential impact on student achievement and are necessary variables for establishing accurate prediction models [9]. Data were collected by means of education management system and questionnaire survey. The education management system provides structured performance data, and the questionnaire survey obtains students’ individual characteristics and learning behavior data to ensure the comprehensiveness and accuracy of the data. Integrate multi-dimensional data to build a multi-level and multi-angle database to support subsequent analysis and modeling.

Sample selection criteria

The samples come from multiple grades and disciplines of a university, including science and engineering, liberal arts and art majors, covering students of different disciplines [10]. The selected students must have at least one semester of learning records during the data collection period to ensure the integrity and reliability of the performance data. In the selection of individual factors, the samples were from students with different family backgrounds, including variables such as different income levels and parents’ education levels, so as to avoid sample bias. In order to ensure the representativeness of the data, the sample covers a certain number of students with high, medium and low scores, and reflects the characteristics of students with different learning levels in the model [11]. The basic information of the sample is shown in Table 1 below.

Table 1. Basic information of samples

Sample ID	Subject category	Grade Level	Gender	Family income	Parental education level	Final score
1	Science & Engineering	Year 1	Male	5000	Bachelor’s degree	85
2	Humanities	Year 2	Female	8000	Master’s degree	78
3	Arts	Year 3	Male	4000	High school	65
4	Science & Engineering	Year 2	Female	6000	Bachelor’s degree	92
5	Humanities	Year 1	Male	5500	High school	72

Data processing methods

The original data is cleaned, missing values and outliers are removed, and mean filling and nearest neighbor filling are adopted to ensure data integrity. The outliers were identified and eliminated by standard deviation method to avoid interference with the analysis results [12]. In order to eliminate the influence between different dimensions, all data are standardized and converted to a uniform scale for subsequent analysis. The Z-score standardization method is used to convert each variable into a standardized value with a mean of 0 and a standard deviation of 1 to ensure comparability between different data sources. In order to improve the utilization efficiency of data, feature engineering technology is used to reduce and extract features from the original data, screen out the feature variables that are strongly correlated with students’ scores, and improve the accuracy of the prediction model [13]. In the data processing process, all steps are carried out in strict accordance with the data analysis specifications to ensure the quality and reliability of the data.

Model construction

Model selection

This study combines fuzzy logic and data mining technology to select common prediction models for comparison. Examples include decision trees, support vector machines (SVM), artificial neural networks (ANN), and fuzzy decision support systems (FDSS). As shown in Table 2 below, the decision tree model makes decisions through the tree structure, is sensitive to data noise, and cannot deal with complex fuzziness [14]. SVM can classify and regression in high dimensional space, and has strong generalization ability. It is not as flexible as fuzzy logic method to deal with nonlinear relation. Artificial neural networks have excellent performance in dealing with nonlinear problems, and the training of large-scale data sets requires relatively high computational resources [15]. Compared with these models, fuzzy DSS can process fuzzy data, adapt to the fuzzy relationship among various factors, and has strong explanatory ability. Therefore, this study finally chooses fuzzy DSS as the main prediction model. The input variables for the fuzzy decision support system (FDSS) model include several key factors: students’ historical achievements (e.g., past grades), learning behaviors (e.g., study time, homework completion), class participation, family background (e.g., parental education level and family income), and mental health status. These variables are considered to have a direct or indirect impact on student achievement. The expected output of the FDSS model is a prediction of a student’s final grade or academic performance, which is calculated based on the fuzzy rules derived from the input variables. These outputs provide insights into how students might perform in future academic tasks, helping educators tailor personalized learning strategies. The input variables for the fuzzy decision support system (FDSS) model include several key factors: students’ historical achievements (e.g., past grades), learning behaviors (e.g., study time, homework completion), class participation, family background (e.g., parental education level and family income), and mental health status. These variables are considered to have a direct or indirect impact on student achievement. The expected output of the FDSS model is a prediction of a student’s final grade or academic performance, which is calculated based on the fuzzy rules derived from the input variables. These outputs provide insights into how students might perform in future academic tasks, helping educators tailor personalized learning strategies.

Table 2. Comparison of different models

Model	Advantages	Disadvantages	Applicable Scenarios
Decision tree	Easy to understand and implement, handles nonlinear data	Sensitive to noise, prone to overfitting	Medium-sized datasets
Support vector machine (SVM)	Good generalization ability, suitable for high-dimensional data	High computational cost for large datasets	Small datasets, classification tasks
Artificial neural networks (ANN)	Excellent at modeling non-linear relationships	High resource consumption, difficult to interpret	Non-linear problems, complex tasks
Fuzzy decision support system (FDSS)	Handles fuzzy data effectively, strong interchangeability	Complex training process, parameter selection challenges	Fuzzy data handling, complex decision tasks

Model architecture design

The architecture design of fuzzy decision support system (FDSS) is based on the fuzzy reasoning mechanism, combining the characteristics of student achievement prediction, fuzzing the input data, building the fuzzy rule base, making inference decision, and finally outputting the prediction result. The core components of fuzzy decision support system include input layer, fuzzy processing module, inference engine, denitrification module and output layer [16]. In the input layer, students’ performance data and influencing factors (such as learning time, family background, etc.) are blurred and converted into fuzzy values. The fuzzy data is entered into the inference engine, and the inference is carried out through the preset fuzzy gauge to generate the intermediate result [17]. The inference engine computes the relationship between the input variable and the output variable according to the fuzzy rule base. The denitrification module converts the inference result into a specific prediction result and outputs it to the result layer. The core formula of fuzzy reasoning is as follows:

y = \sum_{i = 1}^{n} w_{i} \cdot R_{i} \cdot f (x_{i})

Y is the prediction result, $w_{i}$ is the weight of the fuzzy rule in Article i, $R_{i}$ is the membership degree of the fuzzy rule, and $f (x_{i})$ is the fuzzy value of the input variable $x_{i}$ in the rule base. The fuzzy rules used in the model are based on the relationships between the input variables and the output (student achievement). For example, if a student spends more time studying and participates actively in class, the system might use a rule like: “IF study_time IS high AND class_participation IS active THEN achievement IS high.” These rules are fuzzy in nature, meaning that the input variables (study time, participation) do not have precise values but instead are represented as fuzzy sets (e.g., “high,” “medium,” “low”). The fuzzy inference engine processes these rules and generates a prediction about the student’s achievement. Equation (1) represents the fuzzy reasoning mechanism, where the prediction result is calculated based on the membership degrees of the fuzzy values and the weights of the rules.

Key feature selection method

In this study, a feature selection method based on information gain and correlation analysis is adopted. The information gain is sufficient to measure the contribution of each feature to the target variable (i.e., student achievement), and the feature with greater information gain is selected as the input of the model. Correlation analysis determines the correlation between features, eliminates highly correlated redundant features, reduces the complexity of the model, and prevents over fitting [18]. In the process of feature selection, learning time, class participation, historical achievements, family background and other core features are selected in combination with domain knowledge. In the process of feature selection, information gain is calculated as follows:

I G (D, A) = H (D) - \sum_{v \in A} \frac{| D_{v} |}{| D |} H (D_{v})

IG(D, A) is the information gain of feature A on data set D, H(D) is the entropy of data set D, Av is the subset of data whose value is v on feature A and H(Av) is the entropy of this subset. Information gain is used to select the most relevant features from the data by measuring how much a specific feature helps reduce uncertainty in predicting student achievement. Features that result in higher information gain are considered more important for the prediction model and are selected for further analysis.

System implementation process

System implementation includes data multiprocessing, model training, result prediction and system evaluation. In the data multiprocessing stage, all input data are cleaned, standardized and feature selected [19]. The fuzzy decision support system will train the model according to the processed data, adjust the fuzzy rules and weights, and continuously optimize the prediction results. In the prediction stage, the system obtains the prediction result through fuzzy reasoning based on the new input data, performs DE-calcification processing, and outputs the final result [20]. In the stage of system evaluation, cross-validation method was used to evaluate the model, and the prediction model was calculated as follows:

\hat{y} = \frac{\sum_{i = 1}^{n} w_{i} \cdot f (x_{i})}{\sum_{i = 1}^{n} w_{i}}

$\hat{y}$ is the predicted student score, $w_{i}$ is the weight of the fuzzy rule, and $f (x_{i})$ is the fuzzy value of the input feature in the fuzzy rule.

Model training and optimization

Division of training set and verification set

In order to ensure robust model training and evaluation, the dataset was divided into a training set and a validation set. The basic standard for splitting the data was as follows:

Training Set: Typically, 70% of the total dataset was used for training the model. This set of data is used to fit the model and adjust its parameters.
Validation Set: The remaining 30% of the data was used for validating the model’s performance. This set is crucial for evaluating how well the trained model generalizes to unseen data and helps in assessing the accuracy of predictions.
Cross-Validation: To avoid bias in the dataset partitioning and ensure that the model is evaluated on multiple splits of the data, cross-validation was applied. In this method, the dataset is split into multiple subsets, and the model is trained and validated several times, each time using a different fold as the validation set while the rest of the data is used for training. This procedure helps in minimizing over fitting and improving the generalization capability of the model.
Randomization: Before splitting the dataset, it was randomized to ensure that each subset represents the overall distribution of the data, preventing any unintended bias that could occur from the order in which data points are presented.

The quality of model training depends on the partitioning of data sets. In this study, cross-validation method is used to partition data sets. The data sets will be divided into training sets and validation sets. The training set is used to train the model, and the validation set is used to evaluate the predictive performance of the model. To avoid data bias, the ratio of training set to validation set is generally set to 70% and 30%. In practice, cross-validation divides the data set into different training sets and validation sets many times to enhance the generalization ability of the model. After each training, the performance of the model on the verification set is evaluated, and the model parameters are adjusted according to the results. Multiple iterations of training and validation to select the best performing model configuration.

Selection of optimization algorithm

In order to improve the prediction accuracy of the model, genetic algorithm, particle swarm optimization (PSO) and gradient descent method are selected. When selecting optimization algorithm, considering the characteristics of fuzzy decision support system, particle swarm optimization algorithm is considered to be more suitable to deal with multi-objective and multi-variable optimization problems. Particle swarm optimization simulates the process of bird feeding and finds the optimal solution in high dimensional space. Compared with other algorithms, particle swarm optimization algorithm is faster and can avoid falling into local optimal solutions. In this paper, particle swarm optimization algorithm is used to optimize the parameter configuration of fuzzy decision support system and improve the prediction accuracy. The comparison of different optimization algorithms is shown in Table 3 below.

Table 3. Optimization algorithm parameter settings

Algorithm	Parameter	Value
Particle swarm optimization (PSO)	Inertia weight (w)	0.7
	Cognitive coefficient (c₁)	1.4
	Social coefficient (c₂)	1.4
	Maximum iterations	500
Gradient descent	Learning rate (η)	0.01
Gradient descent	Maximum iterations	1000
Genetic algorithm	Population size	50
	Mutation rate	0.05
	Crossover rate	0.8

The update of the particle swarm optimization algorithm is as follows

v_{i}^{t + 1} = w \cdot v_{i}^{t} + c_{1} \cdot r_{1} \cdot (p_{i}^{t} - x_{i}^{t}) + c_{2} \cdot r_{2} \cdot (g^{t} - x_{i}^{t}) x_{i}^{t + 1} = x_{i}^{t} + v_{i}^{t + 1}

The optimization algorithms play a crucial role in fine-tuning the parameters of the FDSS model to improve its prediction accuracy. In this study, Particle Swarm Optimization (PSO) was chosen over Genetic Algorithms (GA) for several key reasons:

PSO vs. GA: PSO is generally preferred for problems with continuous search spaces, where solutions can be represented as real-valued vectors. PSO simulates the social behavior of birds flocking or fish schooling to explore the solution space. This algorithm is known for its ability to converge quickly and avoid local minima, which can sometimes be a limitation of GA. In contrast, Genetic Algorithms are typically more suited for problems with discrete variables and rely on genetic operators such as crossover and mutation to evolve the population.

Efficiency: PSO is computationally more efficient in terms of convergence speed and solution quality, especially in continuous optimization problems like parameter tuning in FDSS. The particle-based search process allows the algorithm to explore the solution space in parallel, significantly improving its search efficiency compared to GA, which is more dependent on crossover and mutation steps that require more computational effort.

Scalability: PSO tends to perform better in high-dimensional spaces, making it more calculable when dealing with a large number of parameters to be optimized.

Application scheme

System deployment scheme

In order to ensure the efficient operation and stability of the fuzzy decision support system, the system deployment scheme involves the choice of hardware and software environment, the design of system architecture and the technical requirements in the implementation process. The hardware configuration of the deployment environment should support large-scale data processing to ensure fast response for model training and prediction. The system needs to be run on a high-performance server, and it is recommended to be equipped with a multi-core processor and enough memory to handle a large student data set. In terms of software environment, the system development adopts mainstream programming languages such as Python and Java, and uses corresponding machine learning frameworks (such as TensorFlow, kitsch-learn) and database management systems (such as MySQL, PostgreSQL) to store and query student data. The system should support a distributed computing architecture that allows tasks to be distributed among multiple nodes to improve computing efficiency. During the deployment process, ensure that the system is calculable, so that it is easy to add more functional modules or handle larger data sets in the future.

Application of grade prediction and analysis

The fuzzy decision support model predicts students’ grades, analyzes the prediction results, and makes real-time prediction based on the trained model after inputting students’ relevant information (such as learning time, historical grades, class participation, etc.). Predictive outcomes provide an estimate of a student’s final grade, providing personalized learning guidance for teachers. The system also has an analytical function to predict trends based on students’ historical achievement and behavior data, which helps education administrators identify students with higher academic risk and take appropriate intervention measures. The predicted results are shown in Fig. 1 below.

Fig. 1 [Images not available. See PDF.]

Comparison of predicted results

Strategies for improving learning effect

The system proposes customized learning plan according to students’ historical achievement and predicted result, combined with learning behavior and mental state. For students with lower grades, the system may suggest increasing study time, adjusting study methods, or scheduling remedial classes. For students with higher grades, extension courses and challenging tasks can be provided to enhance their learning ability. Based on group data, the system can also identify groups of students who have difficulties in specific subjects or learning areas, and propose targeted collective intervention measures, such as group discussions and tutorial classes. The combination of individuation and group strategy can maximize the learning effect of students and optimize the allocation of educational resources.

Results and discussion

Results

Forecast results

The fuzzy Decision support system (FDSS) is used to predict student achievement, and different samples are tested. The system can accurately predict students’ grades according to the input data. The error between the predicted result and the actual result is small in the analysis of students’ historical achievement, learning behavior, family background and other factors. In the test sample, the prediction error of the system for most of the students is kept within 5%, showing a good prediction accuracy. As shown in Fig. 2 below, the overall distribution of prediction errors is concentrated, with most students’ grade prediction errors ranging from 2 to 3%. The results show that the fuzzy decision support system is effective in the task of student achievement prediction, especially in the complex multi-variable environment, the model can estimate the student achievement more accurately. The x-axis represents the range of prediction errors, and the y-axis shows the frequency of these errors across the test samples. The majority of the errors are between 2 and 3%, indicating the accuracy of the FDSS model in predicting student achievement.

Fig. 2 [Images not available. See PDF.]

Prediction results

Comparative analysis

Compared with the traditional regression model, decision tree model and support vector machine (SVM) model, it is found that FDSS is superior to other models in prediction accuracy and generalization ability. The average prediction error of FDSS is significantly lower than that of other models, which can better adapt to the fuzziness of input data. As shown in Fig. 3 below, the accuracy of FDSS model is higher than that of other traditional prediction models, and the error is smaller. Although the decision tree and SVM models perform well in some cases, the prediction error and accuracy are still lower than the FDSS. This figure compares the prediction accuracy of FDSS with traditional models such as decision trees, support vector machines (SVM), and neural networks. The FDSS model shows superior accuracy with lower prediction errors compared to the other models.

Fig. 3 [Images not available. See PDF.]

Prediction accuracy statistics

Model performance evaluation

In this study, the performance of the FDSS model is evaluated in multiple dimensions, and the evaluation results show that the FDSS model has excellent performance in all performance indicators, and has achieved good results in accuracy and F1 scores. In the complex task of student achievement prediction, FDSS can deal with the fuzzy relationship between multiple variables effectively and provide reliable prediction results. As shown in Fig. 4 below, the FDSS model has an accuracy of 97.85%, an accuracy of 96.62%, a recall of 98.73%, and an F1 score of 97.67%. FDSS not only achieves high accuracy in student achievement prediction, but also provides consistent predictions at all levels of identifying student achievement levels. This figure displays the evaluation metrics for the FDSS model, including accuracy, recall, and F1 score. The FDSS model achieved high performance with an accuracy of 97.85%, recall of 98.73%, and F1 score of 97.67%.

Fig. 4 [Images not available. See PDF.]

Results of model performance evaluation indicators

Application effects of the model on different data sets

This paper applies fuzzy decision support system to different student data sets, such as student achievement data of different subjects, different grades and different regions. The performance of FDSS model on all kinds of data sets is relatively stable, and it can adapt to different student groups and subject characteristics. FDSS is able to provide accurate performance prediction for all subjects, whether in science or engineering, arts or arts. As shown in Fig. 5 below, the FDSS model performs well across datasets of different disciplines. In the data set of science and engineering, the accuracy of the model reaches 98.12%. In liberal arts and art subjects, the accuracy rate has decreased slightly, still staying above 95%, which proves the wide adaptability and strong generalization ability of FDSS model. This figure shows the prediction accuracy of the FDSS model applied to different subject areas (Science, Engineering, Humanities, Arts). The model performs well across all subjects, with accuracy rates above 95%.

Fig. 5 [Images not available. See PDF.]

Application effect

Discussion

Problem summary

In this paper, a student achievement prediction model based on fuzzy Decision support system (FDSS) is constructed to explore the accuracy and feasibility of student achievement prediction. The experimental results show that FDSS model has high prediction accuracy and good generalization ability, and runs stably on different data sets and disciplines. Although the FDSS model performs well in most tests, there are still problems that need to be explored and solved. The model’s performance is relatively weak when dealing with extreme values. There are large errors in the scores of some samples. When there are significant outliers in the input data, the accuracy of the prediction results will be affected to some extent. The current model may not be sufficiently robust when dealing with abnormal data, especially in groups of students with very low or very high grades. The feature selection process has achieved good results and can still be optimized. The model mainly relies on students’ historical achievement, learning behavior and other conventional characteristics to predict, and ignores some potential factors, such as psychological state and family education environment. These factors have an impact on students’ performance in some cases. The computational complexity and training time of the model are challenging when dealing with large-scale data. In this study, optimization methods such as particle swarm optimization algorithm are adopted. As the amount of data increases, the training and prediction time of the model is still long.

The potential scalability of the FDSS model extends beyond higher education into other educational environments, such as K-12 education and vocational training. The model’s core ability to handle multi-dimensional data and predict outcomes based on uncertain and fuzzy inputs makes it adaptable to a variety of educational contexts. In K-12 education, for example, the FDSS model could be used to predict student achievement not only based on academic performance but also on additional factors such as classroom behavior, emotional states, and extracurricular activities. As K-12 students have more varied learning environments, the flexibility of the FDSS in incorporating different types of data is particularly advantageous.

In vocational training, the FDSS model could be employed to predict trainee performance based on factors such as prior work experience, skills development, and participation in practical training sessions. The model’s fuzzy reasoning mechanism would allow it to handle the uncertainties inherent in such environments, where traditional metrics like exam scores may not fully capture the breadth of a trainee’s competencies. Furthermore, by integrating data on job placement rates and post-training success, the model could provide predictions on career outcomes for vocational trainees, offering valuable insights for curriculum design and training program effectiveness.

Research suggestions

Strengthen the processing ability of outliers, filter and correct outliers through the data multiprocessing stage, and introduce more robust algorithms to improve the performance of the model in abnormal data. Researchers should also consider using more complex data multiprocessing techniques, such as data standardization and normalization, to improve the adaptability of the model to various data distributions. In order to improve feature selection, in addition to the existing performance and learning behavior data, future research attempts to comprehensively model students’ psychological and emotional states, family background and other factors. Helps provide more accurate grade prediction results in high-risk student populations. By means of questionnaire survey and psychological test, relevant data are collected and introduced into the model for multidimensional analysis. Improve the computational efficiency and scalability of the model, for the processing of large-scale data sets, explore distributed computing and parallel computing techniques, and improve the speed of model training and prediction. The use of cloud computing platforms to provide computing resources will provide more flexible support for large-scale data analysis.

Conclusion

In this paper, a student achievement prediction model based on fuzzy Decision support system (FDSS) is proposed, and its effectiveness is verified by experiments. The results show that FDSS model can integrate the multi-dimensional characteristic data of students well and provide accurate grade prediction when dealing with the task of student achievement prediction. Compared with different models, FDSS has better predictive performance than traditional machine learning methods (such as decision trees, support vector machines, etc.). In terms of application, FDSS can predict the performance of individual students, provide comprehensive analysis reports for education administrators, help identify students with high academic risk, and develop personalized learning plans. The stability and adaptability of FDSS in different disciplines and student groups provide strong support for educational decision-making. The study also exposed problems with the model’s weak robustness in handling outliers, the fact that the dimensions of feature selection can still be expanded, and the computational efficiency of processing large-scale data sets. In general, the student achievement prediction model based on fuzzy decision support system provides a new idea and method for intelligent decision making in the field of education. With the development of data science and artificial intelligence technologies, similar predictive models will become more accurate and efficient, providing educational administrators and teachers with more comprehensive support for student assessment.

Author contributions

X.L. wrote and reviewed this manuscript.

Funding

No funding received.

Data availability

The datasets used during the current study available from the corresponding author on reasonable request.

Declarations

Ethics approval

The protocol was approved by Institute of Education-Industry Synergy for Talent Development of Sanya University of the ethics committee.

Informed consent

The informed consent to participate was obtained from the participants.

Competing interests

The authors declare no competing interests.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Abdasalam, M; Alzubi, A; Iyiola, K. Student grade prediction for effective learning approaches using the optimized ensemble deep neural network. Educ Inf Technol; 2024; [DOI: https://dx.doi.org/10.1007/s10639-024-13224-7]

2. Govindan, AR; Li, XM. Fuzzy logic-based decision support system for automating ergonomics risk assessments. Int J Ind Ergon; 2023; 96, [DOI: https://dx.doi.org/10.1016/j.ergon.2023.103459] 103459.

3. Lee, S; Lee, JH; Jeong, Y. The effects of digital textbooks on students’ academic performance, academic interest, and learning skills. J Mark Res; 2023; 60, 4 pp. 792-811. [DOI: https://dx.doi.org/10.1177/00222437221130712]

4. Xiao, JE; Teng, HQ; Wang, H; Tan, JX. Psychological emotions-based online learning grade prediction via BP neural network. Front Psychol; 2022; 13, [DOI: https://dx.doi.org/10.3389/fpsyg.2022.981561] 981561.

5. Hoyos, W; Aguilar, J; Toro, M. A clinical decision-support system for dengue based on fuzzy cognitive maps. Health Care Manag Sci; 2022; 25, 4 pp. 666-681. [DOI: https://dx.doi.org/10.1007/s10729-022-09611-6]

6. Burda, M; Gojová, A; Grundelová, B; Malina, M; Stanková, Z; Stepnicka, M; Vajgl, M. EVKA-fuzzy modelling based system for the decision-making support of community workers. Soc Sci Comput Rev; 2023; 41, 2 pp. 648-673. [DOI: https://dx.doi.org/10.1177/08944393221117912]

7. Nabizadeh, AH; Goncalves, D; Gama, S; Jorge, J. Early Prediction of students’ final grades in a gamified course. IEEE Trans Learn Technol; 2022; 15, 3 pp. 311-325. [DOI: https://dx.doi.org/10.1109/TLT.2022.3170494]

8. Garg, RK; Garg, R. Decision support system for evaluation and ranking of robots using hybrid approach. IEEE Trans Eng Manage; 2023; 70, 9 pp. 3283-3296. [DOI: https://dx.doi.org/10.1109/TEM.2021.3079704]

9. Alavi, B; Tavana, M; Mina, H. A dynamic decision support system for sustainable supplier selection in circular economy. Sustain Prod Consum; 2021; 27, pp. 905-920. [DOI: https://dx.doi.org/10.1016/j.spc.2021.02.015]

10. Jankova, Z; Jana, DK; Dostal, P. Investment decision support based on interval type-2 fuzzy expert system. Eng Econ; 2021; 32, 2 pp. 118-129. [DOI: https://dx.doi.org/10.5755/j01.ee.32.2.24884]

11. González, MD; Talavera-Velasco, B; Gutiérrez, SU. The role of engagement and temporal perspective in the academic performance of postgraduate students. Physiol Behav; 2020; 224, [DOI: https://dx.doi.org/10.1016/j.physbeh.2020.113054] 113054.

12. McDermott, AF; Rose, M; Norris, T; Gordon, E. A novel feed-forward modeling system leads to sustained improvements in attention and academic performance. J Atten Disord; 2020; 24, 10 pp. 1443-1456. [DOI: https://dx.doi.org/10.1177/1087054715623044]

13. Arnaiz-Sánchez, P; de Haro, R; Alcaraz, S; Ruiz, ABM. Schools that promote the improvement of academic performance and the success of all students. Front Psychol; 2020; 10, 2920. [DOI: https://dx.doi.org/10.3389/fpsyg.2019.02920]

14. Willis, A; Hyde, M; Black, A. Juggling with both hands tied behind my back: teachers’ views and experiences of the tensions between student well-being concerns and academic performance improvement agendas. Am Educ Res J; 2019; 56, 6 pp. 2644-2673. [DOI: https://dx.doi.org/10.3102/0002831219849877]

15. Lemay, DJ; Doleck, T. Grade prediction of weekly assignments in MOOCs: mining video-viewing behavior. Educ Inf Technol; 2020; 25, 2 pp. 1333-1342. [DOI: https://dx.doi.org/10.1007/s10639-019-10022-4]

16. Hilbert, S; Bruckmaier, G; Binder, K; Krauss, S; Bühner, M. Prediction of elementary mathematics grades by cognitive abilities. Eur J Psychol Educ; 2019; 34, 3 pp. 665-683. [DOI: https://dx.doi.org/10.1007/s10212-018-0394-9]

17. Sremac, S; Zavadskas, EK; Matic, B; Kopic, M; Stevic, Z. Neuro-fuzzy inference systems approach to decision support system for economic order quantity. Econ Res Ekon Istraž; 2019; 32, 1 pp. 1114-1137. [DOI: https://dx.doi.org/10.1080/1331677X.2019.1613249]

18. Allen, J; Mattern, K. Examination of indices of high school performance based on the graded response model. Educ Meas Issues Pract; 2019; 38, 2 pp. 41-52. [DOI: https://dx.doi.org/10.1111/emip.12250]

19. Nobari, SM; Yousefi, V; Mehrabanfar, E; Jahanikia, AH; Khadivi, AM. Development of a complementary fuzzy decision support system for employees’ performance evaluation. Econ Res Ekon Istraž; 2019; 32, 1 pp. 492-509. [DOI: https://dx.doi.org/10.1080/1331677X.2018.1556106]

20. Vernon-Feagans, L; Mokrova, IL; Carr, RC; Garrett-Peters, PT; Burchinal, MR; Greenberg, MT; Blair, CB; Cox, M; Frank, JL; Mills-Koonce, WR; Willoughby, MT. Cumulative years of classroom quality from kindergarten to third grade: prediction to children’s third grade literacy skills. Early Child Res Q; 2019; 47, pp. 531-540. [DOI: https://dx.doi.org/10.1016/j.ecresq.2018.06.005]

Word count: 5460

Show less

Student achievement prediction and auxiliary improvement method based on fuzzy decision support system

Content area

Abstract

Full text