This research-based project titled "Predicting Stress Level in Online Education Due to Coronavirus Pandemic: A Case Study of Bangladesh Students", submitted by Sunzida Siddique Sajal Baidya to the Department of Computer Science and Engineering, Daffodil International University has been accepted as satisfactory for partial fulfillment of the requirements for the degree of B.Sc. Touhid Bhuiyan, Professor and Head, Department of CSE and also to other faculty members and the staff of the CSE Department of Daffodil International University. We would like to thank all our coursemates at Daffodil International University who participated in this discussion while completing the coursework.
To predict this problem, the accuracy was 74 percent given by Random forest by selecting all features and 66 percent given by Xgb Boost by selecting Limited feature. We used Random Forest Classification Algorithms, Correlation Heatmap and Univariate Selection to select necessary data and to remove the unnecessary data not related to the prediction of students' stress level in online classes discussed in a separate section of our paper not.
INTRODUCTION
- Introduction
- Motivation
- Rationale of The Study
- Research Question
- Expected Output
- Project Management and Finance
- Report Layout
After that, even though the online class education system was introduced, the students faced various pressures. Not everyone can continue the education system equally and many educational institutions do not have the capacity to continue the online education system. Along with all the countries of the world, Bangladesh also has problems in getting people.
Corona has had a huge impact on the education system of Bangladesh along with the rest of the world. However, most countries are poor and not everyone has the opportunity to study in a good educational institution and not all educational institutions have the capacity to continue their education online. The financial system of Bangladesh has suffered greatly and the education system has suffered greatly.
For this corona situation, we will be able to find out through our work how the corona has affected the education system of Bangladesh and what kind of emotional change has happened among the students as a result of holding online classes. By determining these activities, we will be able to measure the problems students face in the online education system during the Covid-19 pandemic in Bangladesh.
BACKGROUND
Terminologies
Nearest Neighbors (KNN)
- Related Works
- Comparative Analysis and Summary
- Scope of The Problem
- Challenges
It helps a lot to reduce loss by helping to avoid overfitting a model built and developed solely for model performance and computational speed. Shweta [1] discusses online classes versus offline classes that are more relevant to the student. This article focuses on grade point average with an average grade of 6.00 in the online class and 5.76 in the offline class.
As a result of the survey, 10.3% of students gave online education as positive feedback and 71.45% of students gave negative feedback. Data was collected from UNESCO where 871 percent from 160 countries have the total student registration. Kruskal Wallis test, Fisher's exact test and Pearson Chi Square test were used in this paper with a statistical value of P < 0.05 was.
This paper suggests that sample testing is used in this paper and work is done with a specific area. From the comparative analysis in Table 2, we see that a lot of research has been done on the impact of Covid-19 on education and online education, but the performance can be better achieved if algorithms are used together. All educational institutions in Bangladesh, including the rest of the world, were closed during the operation.
Our paper addresses the mental state of students during their online classes in Bangladesh, their opinions and the emotional stress students face when it comes to online classes. In the context of Bangladesh, it has been worked on the mental state of students for online classes during the Corona period and the mental state of students due to the closure of educational institutions in Corona. No such secret has been found and even if such things were found, there is not much research on online classes and pressure to study at this time.
Some studies have been conducted, but they have used data sets from the Internet and may not be valid. Opinions have been taken from students in Bangladesh and their opinions have been taken. Our research has had to face a number of challenges, there have been a number of problems, they have overcome these problems.
RESEARCH METHODOLOGY
Research Subject and Instrument
Data Collection Procedure
Here we have 15 input columns and 1 output column where we have labeled the activity name. Then we have converted it to csv files for one to get our complete dataset. Then we had made Label encoding as our two types of data in our dataset, one numeric and one categorical.
After that, we have split our data set (train set and test set) in the ratio 8:2. After calculating all the features, we use the selected Feature which is found by random forest classifier, correlation and univariate feature selection in Method-2. By mixing and selecting all these 7 common features, we again implemented Machine Learning Algorithm using Restricted Feature.
The complete data collection, storage and preprocessing procedure is shown below in Figure 3.3.
Statistical Analysis
Pearson Correlation: Calculating the covariance of two variables divided by the product of their standard deviations yields the Pearson correlation. The test statistic Pearson's correlation coefficient measures the statistical relationship, or association, between two continuous variables. It provides data on the magnitude of the association, or correlation, as well as the relationship's direction [10]. The formula of Pearson correlation coefficient is given below in Equation 1.
Scikit-learn exposes feature selection routines as objects that implement the transformation method: SelectKBest [11] removes all but the highest-scoring features. We use the f_classif technique for our proposal. The reason for this is that random forest tree strategies are naturally ranked by how well they improve node purity. From the comparison visualization above, each activity we worked on is among the top three most common features.
So we can clearly say that this is our final feature that we select from the similarities of three features.
Proposed Methodology
In this case the error rate of KNN is 0.36, it means if we determine the value of k in the form of error rate if the value of k = 5. In this case the error rate of KNN is 0.36, it means if we determine the value of k in the form of error rate as the value of k = 19. In this case the error rate of decision tree is 0.33, it means if we determine the value of n in the form of error rate as the value of n = 5.
In this case the error rate of svm is 0.32, it means if we determine the value of c in the form of error rate if the value of c = 4. In this case the error rate of svm is 0.33, it means if we determine the value of c in the form of error rate as the value of c = 1. In this case the error rate of NN is 0.33, it means if we determine the value of c in the form of error rate as the value of c = 6.
In this case the error rate of NN is 0.35 which means if we define the value of c in the form of the error rate than the value of c = 1.
Implementation Requirements o Operating System - Windows
Experimental Setup
Experimental Results and Analysis
From the table above, we can see that Random Forest Algorithm showed an almost perfect performance, reaching 74% accuracy. From here we can see our best accuracy given by Random Forest Algorithm which is 74 percent accuracy. From table above, we can see that Xgb Boost Algorithm showed an almost perfect performance, reach 66% accuracy.
We have also trained some other algorithm by selecting limited features to test our data set which is given below from “Table-4.8”. In (Method-2) Xgb Augmentation Algorithm has achieved 66% accuracy. By using Dimensionality Reduction (Method-2) the accuracy will be improved in the future.
Discussion
IMPLICATION ON SOCOTY, ENVIRONMEN AND SUSTAINABILITY
- Impact on Society
- Impact on Environment
- Ethical Aspects
- Sustainability Plan
However, we will be able to find out what the impact of the corona period is on education. Thanks to this project we will be able to learn more about the impact on corona time education, about the impact on education. Our project can be made more effective later with more data and the use of different types of applications.
In addition, we work according to the guidelines to make the work more interesting and effective. The model will increase awareness and its effectiveness will be explained to people, so that people will provide more information and our project will come closer.
SUMMARY, CONCLUSION, RECOMMENDATION, AND IMPLICATION FOR FUTURE RESEARCH
Summary of The Study
Conclusion
Here we learned a lot about machine learning and pedicure that will help us do better in the future.
Implication for Further Study
Available: https://www.oecd.org/education/the-impact-of-covid-19-on-education-insights-education-at-a-glance-2020.pdf. Mazza, The likely impact of COVID-19 on education: reflections based on the existing literature and recent international datasets, vol.
