An Integrated Framework for Students' Academic Performance Prediction Based on Affective and Cognitive State Using Mixed Numeric and Categorical Data

[Virtual Presenter] This research proposes an integrated framework for predicting students' academic performance using mixed numeric and categorical data from their affective and cognitive states. The proposed framework combines machine learning algorithms and Bayesian Networks to analyse complex data from multiple sources to predict students' academic performance. It is aimed at finding the best predictor model for predicting students' academic performance for the given dataset. The results show an improvement in accuracy of predictions when compared to existing models. The proposed framework provides an efficient way for predicting students' academic performance from mixed datasets with greater accuracy." In this research a framework has been proposed to make predictions about students' academic performance based on data from their affective and cognitive states. The proposed framework uses machine learning methods and Bayesian Network to analyse complex data from different sources. We are looking to find a model that can accurately predict the academic performance of students based on the given dataset. The results of this research demonstrate a higher accuracy of predictions when compared to existing models. This framework provides an effective and efficient way of making predictions about students' academic performance from mixed data..

[Audio] The research proposed an integrated framework for predicting students' academic performance using mixed numeric and categorical data from their affective and cognitive states. The presentation outline consists of four phases: the overall workflow the introduction the motivation and the literature review. These phases will assist in defining the problem followed by the development of the proposed framework. A conclusion will be drawn based on the results which will be rounded off with some useful references and other relevant publications..

INTRODUCTION. 3/6/2024. 3.

[Audio] Educational data mining is an area of research that intends to use data analysis and computational techniques to find meaningful knowledge from the considerable amount of data present in the academic domain. Through employing data sets and implementing different computational methods we can recognize significant patterns and give significant insights which can be utilized to enhance learning patterns teaching procedures and academic systems. This research proposed an integrated framework for forecasting students' academic performance by making use of a mix of both numeric and categorical data from their cognitive and affective states..

[Audio] Machine learning has become an important part of many aspects of modern life. Rather than being programmed with specific rules algorithms can be used to enable computers to make decisions and predictions. Recommender systems for instance use machine learning to determine user preferences and provide tailored suggestions. This research proposed an integrated framework that uses data from both affective and cognitive states to predict academic performance. Through this framework educational institutions can gain a better insight into their students' academic performance..

[Audio] Education is a vital part of growing both on a personal and professional level. Educational institutions play a pivotal role in fostering potential capacity and knowledge of their students. This research presented a special and ingenious method for forecasting the academic accomplishment of students by combining data from their cognitive and affective states. Through this we are providing them with the correct resources to progress academically and emotionally so that they can pursue their academic and professional goals..

[Audio] Bloom's Taxonomy proposed by Benjamin Samuel Bloom an American Educational Psychologist is a theoretical framework for learning. It divides learning into three domains: Cognitive which focuses on knowledge; Affective which looks at emotions attitudes and behavior; and Psychomotor which focuses on skills. These domains are further divided into six levels of learning providing educators a comprehensive comprehension of the learning process..

MOTIVATION. 3/6/2024. 8.

[Audio] Consequently the current educational environment can be characterized by a relative lack of emphasis on the affective domain of Bloom's Taxonomy. This can result in a neglect of important elements such as emotional intelligence personalized learning and values and ethics education. To address this issue we have proposed an integrated framework that can predict student academic performance by taking into account a mixed set of numerical and categorical data from the student's affective and cognitive states. The aim is to explore the relationship between the two domains thus allowing us to gain a better understanding of the complexities of student performance and evaluate and optimize educational programs more efficiently..

[Audio] The affective domain of the proposed integrated framework focuses on the self growth and development of a student's emotions attitudes and behaviors. Attributes of this domain include emotional happiness self esteem self efficacy self determination interpersonal skills personality and emotional intelligence. Emotional intelligence can enable students to control their emotions attitude and behavior which is essential for their academic performance. This slide provided the theoretical background of this research which helps to give a better understanding of the proposed integrated framework..

[Audio] The research proposes an integrated framework to predict students' academic performance using data from their affective and cognitive states. Cognitive domain attributes such as knowledge and understanding acquired through instruction are considered to be important for predicting academic performance. To decide which cognitive attributes to consider in the framework the research reviews the literature. Additionally it investigates the connection between cognitive domain attributes and academic performance based on a dataset from the target population..

LITERATURE REVIEW. 3/6/2024. 12.

[Audio] This research aims to create an integrated system for predicting students' academic performance with a combination of numerical and categorical data from their emotional and cognitive states. The table summarizes the research papers related to this framework all of which explore the impact of data mining in education to elevate student performance and tackle social problems. There are also papers that recognize the aspects that lead to students quitting and how machine learning can be employed to distinguish those prone to failing early on so preventive actions can be taken. All in all these studies create a complete depiction of how data mining and machine learning models can provide assistance for our students to succeed..

PROBLEM DEFINITION. 3/6/2024. 14.

[Audio] We are proposing a recommender system for predicting student academic performance which combines numeric and categorical data from their affective and cognitive states. We aim to explore how emotional intelligence affects self-esteem self-determination emotional well-being self-efficacy and personality and which of these attributes has the greatest influence on academic performance. If successful this system could help teachers better understand their students' emotional states and provide them with a way to act as emotional mentors to guide their students to academic success..

[Audio] This research project aims to consider emotional intelligence as an important affective attribute and to select Self Esteem Self Determination Self Efficacy Well Being Positive Emotions Negative Emotions and Personality as second-level affective attributes to design a framework indicating the relationships among them with the goal of predicting student's academic performance using mixed numeric and categorical data from their affective and cognitive states..

[Audio] Data from students' systems of performance is collected and then pre-processed to improve accuracy. A combination of supervised and unsupervised methods is used to train the model. The model is validated with a test dataset and the results of the analysis are presented to the user. This proposed an integrated framework for predicting students' academic performance that combines numeric and categorical data of the affective and cognitive states of the students..

[Audio] A scoring method 'Quest_SA' is proposed for closed ended questionnaires using sentiment analysis with polarity value in Phase 1. Sentiment analysis is a way of recognizing and extracting subjective information from text and other forms of data. It permits us to assess the sentiment for a target phrase based on its polarity which is how positive or negative it is. The polarity values can then be converted into numerical scores for evaluation..

[Audio] We see here Sentiment Analysis as a tool used in our project. It is a method of analyzing quality of texts by classifying the emotions and opinions expressed. It can detect the sentiment and emotions behind a piece of text such as joy fear anger and so on. This helps to determine the attitude of the writer towards the subject of the text. Sentiment analysis has become a valuable tool which can be used to predict students' academic performance and make more accurate predictions..

[Audio] Today I will be discussing our methodology of collecting data on student affective and cognitive states. We proposed to use a combination of open-ended and closed-ended questionnaires to gain the most accurate insights on student learning and academic performance. Specifically open-ended questionnaires allow students to freely respond to questions while closed-ended questionnaires require them to choose the most fitting answer from a predefined limited set of options..

[Audio] Our objective in Phase 1 was to create a sentiment analysis model called QUEST-SA which would identify student affective states by assigning a polarity value for each questionnaire scale. This model integrates a set of machine learning algorithms to analyze the data and assign sentiment scores. This makes it possible to more accurately predict students' academic performance..

[Audio] An integrated framework for predicting students' academic performance was examined in this research. Data was obtained from the students' cognitive and affective states using a mix of numerical data and a closed-ended questionnaire. Responses requested topics such as positive and negative emotions self-esteem self-determination self-efficacy emotional happiness personality emotional intelligence and general self-efficacy. Results showed that this integrated approach was more effective in predicting students' academic performance compared to other approaches..

[Audio] An integrated framework has been proposed that takes advantage of numerical and categorical data drawn from students' affective and cognitive states to build a predictive model for their academic performance. The Questionnaire Evaluation With Proposed Symmetric Aggregation (QUEST_SA) method is used to predict the academic accomplishment of students with mixed variables. Affective and cognitive states are surveyed through a questionnaire to collect data and QUEST_SA is then able to pinpoint the most influential covariates that decide students' academic performance in addition to forming an integrated predictive model..

[Audio] An example of a questionnaire developed based on Daniel Goleman's framework of emotional intelligence can be seen in this slide. It contains 15 questions that are answered using a Likert scale of values ranging from “Not at All” “Rarely” ”Sometimes” ”Often” to “Very Often”. The total score of the respondent's answers will determine whether they have high average or low emotional intelligence. To illustrate this the first two questions of the questionnaire are shown. Through such questions it is possible to gauge a student's emotional intelligence and use it to predict their academic performance..

[Audio] An integrated framework is being examined to forecast students' academic performance. This framework takes into account both numeric and categorical data from their affective and cognitive states to create predictions. To assess the data a common evaluation system is initially utilized. This entails allotting a score of one to five with one signifying 'not at all' and five representing 'very often'. These scores are totaled to compute whether the score is within the range of 15 to 34 which is considered low; 35 to 55 which is regarded as medium; or 56 to 75 which is viewed as high. In addition the proposed QUEST-SA evaluation is conducted and bestows a polarity score of -2 to 2 with -2 representing 'not at all' and plus 2 indicating 'very often'. The computation then necessitates summing up the polarity scores to appraise whether the emotional intelligence is low average or high..

[Audio] In this slide we discuss the use of the Positive and Negative Affect Schedule (P-A-N-A-S) to measure emotions. This scale consists of 20 items of which 10 measure positive affect such as excitement and inspired and 10 measure negative affect such as distressed and upset. Each item is rated on a five-point Likert Scale so that the individual can be categorized as being more positive or negative. This is an important tool to measure how one's emotions affect their academic performance..

[Audio] The Standard Evaluation model assigns a numerical value from 1 to 5 to each category in the data from Very Slightly or Not at All to Extremely and the summation of these scores results in a Positive or Negative result. The Proposed QUEST_SA Evaluation works on the basis of assigning a polarity score from -2 to 2 for each category. The summation of these scores yields a Positive or Negative result. This integrated framework offers an effective way to predict students' academic performance by using data from their affective and cognitive states..

[Audio] Rosenberg's Self Esteem scale is an important part of our system. This ten-item measure enables us to gauge the self esteem of our students in a dependable and confirmed way. Every inquiry is replied to utilizing a 4-point Likert scale permitting us to classify the reactions into three classes: high normal and low. For instance one inquiry requests that we rate how we generally feel about ourselves while another inquiry requests that we survey how we at times feel about ourselves. Responding to both of these inquiries gives us an understanding of the student's self-esteem..

[Audio] When predicting students' academic performance two evaluation methods can be used. The first is the standard evaluation which assigns a score of 0-14 for 'strongly agree' 'agree' 'disagree' and 'strongly disagree'. Scores of 0-14 are considered low 14-25 are considered normal and scores higher than 25 are considered high. The second method is the proposed QUEST_SA evaluation where a polarity score is assigned to 'strongly agree' 'agree' 'disagree' and 'strongly disagree'. Positive values are considered high zero is considered normal and negative values are considered low. It is important to keep in mind that both these methods offer unique advantages when it comes to predicting students' academic performance..

[Audio] The Eysenck Personality Questionnaire (E-P-Q--) is a tool for measuring personality traits created by psychologists Hans Jürgen Eysenck and Sybil B G Eysenck. It is composed of 48 Yes/No questions that help calculate a score for Extroversion Neuroticism and Psychoticism. Our research has proposed incorporating this data with numerical and categorical data related to affective and cognitive states to predict a student's academic performance..

[Audio] A proposed evaluation framework is presented in this slide to predict students' academic performance based on mixed numeric and categorical data collected from their affective and cognitive states. To do this a table of scores for eight psychometric attributes – psychotism extroversion neuroticism and lie – has been designed. A standard score of either 1 or 0 and a reversed score of either 0 or 1 is assigned for each attribute in accordance with the PROPOSED QUEST_SA evaluation. Through this a quantitative measure of the influence of students' affective and cognitive states on their academic performance is offered..

[Audio] The Self-Determination Scale is a tool developed by KM Sheldon to assess individual personalities and how they affect behaviour. It is a 10-item likert scale consisting of two five-item subscales each rated from one to five. This tool is based on the concept of individuals being more conscious of their emotions and having the freedom to decide their actions. The S-D-S can be used to foretell a student's academic performance from numeric and categorical data generated from their cognitive and emotional dispositions..

[Audio] We are proposing a two-step evaluation process to use data from a student’s affective and cognitive states to predict their academic performance. The first step is to obtain a standard score by summing the data and dividing it by 10. This will give a low score if the result is under 3 and a high score if the result is over 3. The second step is the proposed QUEST-SA evaluation which uses polarity calculation. A positive result means a high score while a negative result is associated with a low score. Reverse score polarity can be used for data values of 1 3 5 7 and 9..

[Audio] Self-efficacy can be defined as a personal belief in one's capabilities to do something. The General Self-Efficacy Scale is commonly used to measure this through a 10 item psychometric scale. A student's results will then inform whether they have high or low self-efficacy..

[Audio] In this research two techniques for evaluation have been proposed. Standard evaluation calculates the scores which range from 1-4 and then totals them. A score of 20 or higher is seen as high and anything lower than 20 is seen as low. The alternative technique QUEST_SA evaluation assigns polarity scores to data instead. This system assigns -2 to not at all true -1 to hardly true 1 to moderately true and 2 to exactly true. The polarity score is used to determine the score with a positive score indicating a high score and a negative score indicating a low score..

[Audio] The concept of emotional happiness and its relation to predicting student's academic performance will be explored. The Oxford Happiness Inventory consists of 29 items which can be answered using four choices from low to high levels of happiness. Participants' responses will categorize them as either happy moderately happy or unhappy. This data will be used to gauge how it impacts a student's academic performance..

[Audio] The standard evaluation method for predicting students' academic performance is based on a 6-point scale and uses a simple calculation to determine a student's level of happiness. On the other hand the proposed QUEST_SA evaluation method uses a 3-point scale and employs polarity scores to quantify student happiness. Scores of 1 or higher indicate a student is happy while scores of 0 or lower indicate an unhappy student. Both of these evaluation methods can provide insights into a student's academic performance..

PERFORMANCE EVALUATION. 3/6/2024. 38.

[Audio] This research looks into predicting academic performance of students through an integrated framework that considers a range of affective and cognitive data. This includes Emotional Intelligence Self Esteem Self Determination Personality and Self Efficacy. This proposed framework offers an advantage over the standard approach which only uses numeric data and fails to capture a student's affective and cognitive state..

[Audio] This slide looks at the findings of our research when we applied the Quest_SA model to predict the academic performance of students using an integrated framework with mixed numeric and categorical data from their affective and cognitive states. Looking across the table we can see that the model was able to achieve high accuracy and low mean absolute error across all criteria. Emotional intelligence self-esteem positive and negative emotions happiness self-efficacy and self-determination all produced accuracy scores above 94% and mean absolute errors below 20..

[Audio] Proposing the Reclust algorithm we aimed to cluster the gathered data according to different affective and cognitive states. Employing the algorithm provides clusters with significant structure at the same time allowing for efficiency and scalability. Once the clustering had been conducted its validity was checked and it was used to measure the students' academic performance..

[Audio] Data preprocessing is a necessary step for any data analysis task. This presentation proposed an integrated framework for predicting student's academic performance using numeric and categorical data from their affective and cognitive states. The framework included data cleaning to address any missing values and data transformation which involved converting categories to numbers. Converting the data into comparable useful and understandable information creates a dataset which can be used to more accurately predict student performance..

[Audio] An integrated framework was proposed for predicting students' academic performance using numeric and categorical data from their affective and cognitive states. The framework utilized the Criterion Reference Model which is a method of assessing a student's level of attainment against given criteria and expressing the results of the assessment as a correlation between the student's performance and the set criteria. The most usual criterion used to measure academic performance was a student's grade point average (G-P-A--). The G-P-A was then put in one of the five categories: excellent very good good above average or average. Students who scored below 5.0 were deemed to have failed..

[Audio] Clustering analysis is a powerful tool to uncover valuable insights from complex datasets. It is an unsupervised learning algorithm which groups unlabeled data into different clusters. By using clustering analysis on mixed numeric and categorical data from students' affective and cognitive states educators administrators and decision makers are able to make educated decisions which can have an impact on the education system. Such data analysis can provide a more reliable estimation of students' academic performance..

[Audio] Our team proposed an integrated framework for predicting students' academic performance using mixed numeric and categorical data from their affective and cognitive states. We designed an algorithm for clustering both numeric and categorical sources allowing us to identify patterns in the data that could help us better understand student performance and the relationship between affective and cognitive states and academic performance. This enabled us to build a predictive model that could accurately predict student performance..

[Audio] Reclust is a hybrid clustering algorithm which has the capability to process categorical and numerical data. The algorithm follows three stages: initial clustering cluster validation and re-clustering. To perform initial clustering a modified probability and similarity based K-Means (MPS-KM) and modified-Self-organizing Map (M-S-O-M-) are utilized. After initial clustering cluster validation is done which evaluates the clustering result. Re-clustering stage is followed after that which involves re-clustering of the incorrectly clustered data..

Algorithm 1: RECLUST. 3/6/2024. 47.

Reclust : MPS -KM. 3/6/2024. 48.

Algorithm-3 P.isgEu Input: Dataset D = Output: Distance between two categorical For I to A I I. 2. Get unique anyal.ug (AV) 3- For x 1 to AV For y — X+l to AV 4. 5. 6- Forj = I to A s. P I—Compute probability between x and y through j Sum—Sum+P 1 10. End If 11. End For 12- 13. Assign —Sum 14. End For 15- End For 16. End For.

[Audio] Results of the research indicated that the modified Self Organizing Map (M-S-O-M-) algorithm was the most effective in predicting academic performance achieving an accuracy of 97%. Thus the proposed framework provides a powerful tool for predicting student's academic performance and can be used to aid decision-making in educational institutions..