AN EFFECTIVE DATA MINING APPROACH FOR ASSESSING STUDENT’S SATISFACTION IN ONLINE EDUCATION DURING COVID-19 PANDEMIC

algorithm used by using modified data set from Kaggle. Results showed 79.9% precision of education level wise prediction using KNN, 73.7% precision of devices wise prediction using KNN and 88.5% precision of educational level wise predication using SVM, 73.8% precision of device wise prediction using SVM which is showing that the proposed model is significant.

algorithm used by using modified data set from Kaggle.Results showed 79.9% precision of education level wise prediction using KNN, 73.7% precision of devices wise prediction using KNN and 88.5% precision of educational level wise predication using SVM, 73.8% precision of device wise prediction using SVM which is showing that the proposed model is significant.

INTRODUCTION
Due to the COVID-19 pandemic, many countries had to suspend in-person classes and adopt alternative methods of education, such as online education (OE).Several scholarly studies have been conducted to assess the impact of COVID-19 specifically on the teaching system [1].However, further research is needed to understand the effects of this epidemic on the learning process.This emphasizes the importance for Higher Education Institutions (HEIs) to develop comprehensive strategies to meet students' educational standards outside traditional classrooms.To address this issue, [2] explored the role of trust in online educational resources in terms of preparation and various approaches, including mass communication and social interaction programming associated with online education.The researchers also proposed university initiatives to enhance attitudes towards OE, such as providing global education opportunities for both professors and students, enabling them to benefit from online learning (OL).Additionally, it is crucial for colleges and universities to consider potential obstacles to student communication in their decisionmaking processes after an epidemic [3].

EDUCATIONAL DATA MINING
Educational Data Mining (EDM) is an emerging field focused on extracting information from educational data.Given the significant expansion of the sector, EDM has the potential to accurately predict student achievement.Through statistical analysis software techniques, it is possible to make accurate predictions, categorize data, establish connections, and group students effectively.This enables the provision of additional information about potential student dropouts and their performance in various events, thereby enhancing the quality of education.EDM has the potential to benefit not only students but also other stakeholders involved in the educational process.Currently, interactive e-learning technologies and methods facilitate the collection and analysis of student data.However, due to the dynamic nature of data, retrieving this type of information can be challenging [4].

CLASSIFICATION
Classification is one of the oldest analytical techniques and remains widely recognized and commonly employed in data mining.Bayesian classification, which considers the passing grade as a basis, can be utilized in this context.By applying conditionally Bayesian probability, it becomes possible to assess the likelihood of a student being hired by a global corporation when they achieve an "A" grade.Similarly, Bayesian

ONLINE LEARNING DURING THE COVID-19
According to the researchers [9], it was suggested that students who received online education during the COVID-19 pandemic achieved higher academic performance compared to those who did not receive direct instruction from their Higher Education Institution (HEI).The researchers conducted a survey to assess the impact of online learning (OL) on college students during the COVID-19 crisis, analyzing both the positive and negative aspects of OL from the students' perspectives in comparison to traditional learning.Due to the limitations imposed by the COVID-19 outbreak, instructors were compelled to incorporate new teaching methods in order to maintain a satisfactory level of education, which highlighted the excellent adaptability of online learning (OL) in such circumstances [10].

STUDENT PERFORMANCE ANALYSIS
Identifying the appropriate predictors is essential for achieving successful prediction outcomes, as emphasized by [11].When predicting student achievement, it is crucial to acknowledge the factors that impact the dissemination of information.Moreover, numerous educational institutions expressed apprehension regarding the decreasing rates of academic achievement and the frequent occurrence of student withdrawals.

INFLUENCING FACTOR IDENTIFICATION
In their study, the researchers [12] examined the behavioral and educational background of the participants, which encompassed aspects such as behavior, previous assessments, areas for improvement, and prior knowledge of curriculum subjects.They presented a conceptual framework based on the existing literature in their research.

STUDENT BACKGROUND AND BEHAVIOR
In the evaluation of student characteristics [13], a comprehensive analysis of students' progress was conducted, encompassing various aspects such as behavioral patterns, social interactions, and academic background.Among the extensively discussed topics in this analysis was the impact of gender.

STUDENT PERFORMANCE PREDICTION
According to [14], the purpose of grading is to assess the extent to which a student learns and applies the information taught in a course.However, accurately estimating the actual score can be challenging.This challenge has motivated additional efforts by scholars in the field of Educational Data Mining (EDM) to develop effective models for describing student achievement.In addition to providing convenience, prediction plays a crucial role in supporting decision-makers, including teachers, in making appropriate and timely interventions.

RESEARCH METHODOLOGY
Amidst the COVID-19 lockdown, virtually all higher education institutions transitioned from traditional classroom settings to virtual classrooms.In Pakistan, numerous universities initially relied on the Traditional Education System (TES), which posed challenges for students accustomed to TES.However, when the pandemic emerged, academic institutions made the decision to implement online learning environments to save students' time.Additionally, the assessment of students' problem-solving skills during both TES and Virtual Learning Environments (VLE) was a notable topic of concern, as depicted in the provided image.Lastly, an inquiry was made regarding students' perceptions of VLE [15].

PRE-PROCESSING
Following the collection of datasets, cleaning is a key step in obtaining correct data and removing ambiguity from data.It entails several phases such as data cleansing, data processing, feature selection, Additionally, in this study, different professors employed various terms to gather input from their students, such as "loss of online connection" or "no internet."Consequently, it was necessary to preprocess the data to address these variations and ensure consistency in the dataset.preprocessed.

PROPOSED MODEL
The primary objective is to develop a prediction model using machine learning (ML) techniques that can achieve the highest possible accuracy in the SSL system.To accomplish this, two classification approaches, namely K-NN and SVM, are employed.These approaches enable the iterative evaluation of the importance, reliability, and quantity of the extracted features throughout the process.The aim is to optimize the model's performance and enhance the accuracy of predictions.

MACHINE LEARNING TECHNIQUES A) K-Nearest Neighbors (K-NN)
The K-NN approach is a fundamental supervised machine learning technique commonly used in various machine learning tasks.It is favored for its simplicity in comparison to more complex supervised ML methods, making it widely adopted.K-NN is utilized in various fields within the pattern matching paradigm, including image identification, finance, medicine, and forestry [16].

B) Support Vector Machine (SVM)
SVM, a supervised machine learning technique, is effective in addressing both classification and regression problems.Its primary application lies in solving classification challenges.SVM is highly valued in the field of data mining due to its ability to achieve high classification accuracy while utilizing fewer computational resources.

RESULTS AND DISCUSSIONS
This section presents the results of the study.The initial research question explores the impact of virtual classrooms on student achievement.Teachers gathered feedback from students through online forms, and a survey was conducted to assess the challenges faced by students and the outcomes they experienced.The provided chart illustrates the average challenges experienced by BS candidates.The primary outcome indicates that the comprehension of the subject matter was rated at 74%, the readability of the lessons at 76%, internet connectivity issues at 29%, electricity-related challenges at 9%, finding the teaching topics difficult at 8%, and no difficulties reported by 54% of the students.Furthermore, the participation rate during the virtual classroom was recorded at 72%.

PROBLEMS FACED BY MS DURING VIRTUAL LEARNING ENVIRONMENT
The following charts display the average challenges experienced by MS participants during Virtual Learning Environments (VLE).The findings include an analysis of the difficulties encountered by male and female students, providing insights into their respective challenges throughout the VLE.

Fig.4 Problems by MS Students Average
The provided chart presents the average challenges faced by MS candidates.The results indicate that forward-thinking abilities were rated at 80%, presentation readability at 76%, internet connectivity issues at 49%, electricity-related difficulties at 16%, finding the presentation topics challenging by 4% of the students, no difficulties reported by 41% of the students, and a participation rate of 81% during the virtual classroom.

CORRELATION OF UNDERSTANDING OF STUDENTS WITH DIFFERENT FACTORS
Examine the relationship between the students' comprehension and the challenges they encounter inside the VLE.

RESULTS FOR PREDICTIONS
In this section, the author employed a range of machine learning techniques, including SVM and KNN, to delve deeper into the concepts being explored.The author utilized SVM and KNN as part of their investigation and analysis.

Fig.5 Visualizing All Attributes
The applied SVM and KNN on each attribute to explore and predict.[18], the algorithm result was 70.41%.A other compared method that used a computerbased cross-sectional study in the year 2021 [19], the algorithm result is 62%.The other compared method that used Smart PLS in the year 2021 [20], the algorithm result was 55.9%.A other compared method that used the Structural model in the year 2021 [21][22][23][24] result was 64.6%.Thus, the proposed method gives a distinction in achieving the results.

CONCLUSION
The approach involved the collection and analysis of various data sources, including students' feedback and performance, to identify patterns and factors affecting student satisfaction.The results showed that the approach was effective in identifying significant factors that influence student satisfaction, such as course design, instructor support, and course materials.These insights can help educators and administrators to make data-driven decisions to improve the quality of online education and enhance student satisfaction.Furthermore, the approach can be applied in other educational contexts beyond the COVID-19 pandemic and can help to improve online learning experiences in the future.
In summary, this study highlights the importance of using data mining techniques to assess student learning experience for students and help them to achieve their educational goals.
graphs below represent the mean of the issues encountered by BS candidates during VLE.

Fig. 3
Fig.3 Problems by BS Subject Average during VLE.

Table . 1
Correlation of Understanding of Students with Different Factors