International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Student Performance Analyzer Athul Raj V M1, Surya Sushad2, Sneha S3 , Rachana Ramachandran R P4 1Student, Dept.of Information Technology, KMCT College of Engineering, Kerala, India
2Student, Dept.of Information Technology, KMCT College of Engineering, Kerala, India
3Student, Dept.of Information Technology, KMCT College of Engineering, Kerala, India. 4Assistant Professor, Dept.of Information Technology, KMCT College of Engineering,
Kerala, India ---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Education plays a crucial role in shaping
pressure, or inadequate teacher support. These challenges can significantly impede the academic progress of these students, necessitating timely intervention from educators to identify and offer assistance. Predicting students' academic performance enables teachers to identify those in need of additional courses, supplementary assignments, or support services. However, in educational institutions with a sizable student body, analyzing individual student performance can be daunting.
students' future prospects. Additional assignments and projects assigned by instructors can bolster the academic performance of students struggling academically. However, a significant challenge lies in the early identification of students who are at risk. Researchers are actively exploring this issue using Machine Learning techniques. Machine learning finds applications across various domains, including the early identification of at-risk students and providing them with necessary support from instructors. This research delves into the outcomes achieved through Machine Learning algorithms in identifying at-risk students and mitigating student failure. The primary objective of this project is to develop a hybrid model using ensemble stacking methodology to forecast atrisk students accurately. A range of Machine Learning algorithms such as Naïve Bayes, Random Forest, Decision Tree, K-Nearest Neighbors, Support Vector Machine, AdaBoost Classifier, and Logistic Regression are employed. Each algorithm's performance is assessed using diverse metrics, and the study showcases the hybrid model that amalgamates the most effective algorithms for prediction. The model is trained and tested using a dataset containing students' demographic and academic information. Additionally, a web application is designed to facilitate the efficient utilization of the hybrid model and obtain prediction results. The study reveals that employing stratified k-fold cross-validation and hyper parameter optimization techniques enhances the model's performance. Furthermore, the hybrid ensemble model's efficacy is evaluated using two distinct datasets to underscore the significance of data features. In the first combination, utilizing both demographic and academic data, the hybrid model achieves an accuracy of 94.8%. Conversely, when solely academic data is utilized in the second combination, the accuracy of the hybrid model increases to 98.3%. The study's focal point is early prediction of at-risk students, enabling educators to offer timely assistance to students struggling academically.
Early identification of students who might face academic difficulties allows for targeted interventions to improve their prospects before problems escalate. This project primarily focuses on high school students, recognizing the substantial impact of their academic achievements on future educational pursuits. By utilizing a dataset comprising both academic and demographic data, the project aims to assess students' success or failure based on established educational criteria. Success is determined based on year-end average scores, with a passing grade set at 50 or above, and a failing grade below 50. According to educational regulations, students with a year-end general average grade below 50 may progress to the next grade provided they have no more than three failed courses. Identifying potentially struggling students early in the academic year is crucial, though challenging due to the large student population and resource constraints. To address this challenge, various techniques are employed to identify at-risk students, acknowledging the complexities educators face in this endeavor. The project emphasizes the importance of employing differentiated techniques to identify students facing precarious academic situations, particularly in high school settings. A departure from traditional methods, the project aims to develop a hybrid model using machine learning techniques. This hybrid model integrates supervised learning algorithms, including Naive Bayes, Support Vector Machine, Random Forest, Decision Tree, K-Nearest Neighbor, Logistic Regression, and AdaBoost, with the goal of achieving superior predictive outcomes. These algorithms' accuracy is evaluated using diverse metrics to identify the top performers. Ensemble methods such as Bagging, Boosting, and Stacking are employed to combine multiple machine learning techniques, with the stacking ensemble learning approach selected for creating the hybrid model.
Key Words: Ensembling Stacking Method, Machine Learning Techniques, dropout prediction. 1. INTRODUCTION. Throughout the academic term, certain students may encounter obstacles in their studies due to various factors, including psychological issues, familial dynamics, peer
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 2177