A Streamlined Approach to Student Stream Prediction Using an Ensemble Machine Learning Model Rajan Saluja Communications on Applied Nonlinear Analysis, 2025 Finding a stream or course after secondary or senior secondary education is a daunting challenge for students and parents, as numerous options are available in various engineering and non-engineering courses. This decision potentially influences a student’s academic success and career. Most frequently, they take courses with the advice of relatives, neighbors, or career counsellors. Online platforms and Learning Management Systems also exist to offer guidance on stream selection. Still, these systems rely on short-term assessments such as tests, quizzes, or interviews, potentially restricting a student's options. Our research employed the Rajan and Rai (RR) student performance prediction model based on a sophisticated Ensemble Machine Learning approach. Our model incorporates a stack of four multiclass classifiers, namely Decision Tree, k-Nearest Neighbor, Naïve Bayes, and One vs. Rest Support Vector Machine classifiers, and demonstrates a remarkable accuracy rate of 80% for predicting the most suitable academic stream for a student in an Institution. To develop this model, we utilized data from five distinct branches of students. We aim to enhance students' academic success so they can complete their degrees with excellent Grades. Exploring our model in the education sector empowers students with the timely facilities they need for a successful and fulfilling educational journey.
Breast Cancer Detection Using Ensemble Machine Learning Stacking Based Model Rashmi, Rajan Saluja Proceedings of 2nd International Conference on Computational Intelligence and Computing Applications Iccica 2025, 2025 Cancer has emerged as one of the most threatening causes of death in the entire globe and the patients of Breast cancer are growing day in day out throughout the globe. Prediction and identification of the stage of cancer are critical to treatment and saving the life of a woman as soon as possible. The conventional disease diagnostic tools such as the mammography and biopsy offer numerous false outcomes that can leave a mess in any family and the patient may end up with other illnesses such as depression and hyper tension. We have utilized an ensemble machine learning model that was suggested in an earlier research on a different multi-classification to predict and detect the breast cancer in our research work. The stacking based model entails stacking of four classifiers Support Vector Machines, Gaussian Naive Bayes, K-Nearest Neighbours, and Decision Tree. We have trained the model and tested it using the famous cancer dataset available publicly to researchers in Kaggle. It is a model, which is very predictive with a high degree of accuracy, precision, recall and Fl-score of the disease as high as 97 percent. According to the results and findings, the model is better than most of the models and other traditional methods of detecting the disease. The model is proposed in the multi-classification, however, in this study work the model is used in binary classification and yields results as sensible as in multi-classification problems.
Measuring and Analyzing the Time Complexity of a Prediction Model in Different Scenarios Rajan Saluja, Munishwar Rai Ssrg International Journal of Electronics and Communication Engineering, 2024 These days, almost every industry uses machine learning techniques. These techniques improve the accuracy of predicting the target output by using a wide range and velocity of data. The goal of each method is to quickly and accurately predict the target value. In this research, the execution time, which is the total time taken to predict the student’s grades, of the earlier proposed EMLRR model has been calculated. The model is based on an ensemble machine-learning technique: Stacking. Further, we have analyzed the time complexity of the model with other alternatives of stacking by choosing a variety of multiclassification models as meta-models. It has been observed that the proposed model has delivered an accuracy of up to 94% with an execution time of less than 3 seconds. This work uses various platforms, CPUs, and GPUs to analyze the execution time for two different datasets. Various student datasets have been tested to check the model’s efficiency in different scenarios. In addition, a comparative study has been done with other possible combinations of base models by increasing and decreasing the number of base models. The proposed prediction model uses the Stacking of four multiclass models to predict student performance with the best accuracy of up to 94% and 89% for two different student datasets.
Designing new student performance prediction model using ensemble machine learning Rajan Saluja, Munishwar Rai, Rashmi Saluja Journal of Autonomous Intelligence, 2023 Academic success for students in any educational institute is the primary requirement for all stakeholders, i.e., students, teachers, parents, administrators and management, industry, and the environment. Regular feedback from all stakeholders helps higher education institutions (HEIs) rise professionally and academically, yet they must use emerging technologies that can help institutions to grow at a faster pace. Early prediction of students’ success using trending artificial intelligence technologies like machine learning, early finding of at-risk students, and predicting a suitable branch or course can help both management and students improve their academics. In our work, we have proposed a new student performance prediction model in which we have used ensemble machine learning with stacking of four multi-class classifiers, decision tree, k-nearest neighbor, Naïve Bayes, and One vs. Rest support vector machine classifiers. The proposed model predicts the final grade of a student at the earliest possible time and the suitable stream for a new student. A student dataset of over a thousand students from five different branches of an engineering institute has been taken to test the results. The proposed model compares the four-machine learning (ML) techniques being used and predicts the final grade with an accuracy of 93%.
Analysis of Existing ML Techniques for Students Success Prediction Rajan Saluja, Munishwar Rai Pdgc 2022 2022 7th International Conference on Parallel Distributed and Grid Computing, 2022 Providing good academic environment, infrastructure and various learning facilities is not sufficient for Higher Educational Institutions in this competitive era. HIEs have to adopt more trending technologies that can help students to achieve best professional growth. Students’ performance prediction in advance using Supervised and Unsupervised Machine Learning techniques is very much trending for learning contexts in HIEs as it helps administrators to design strategies for improving final results. This is possible as high-volume data about student’s personal, academic performance at school level and at the admission time is available in school, college and universities. Historical data can be utilized to analyze various learning capabilities of different students in different environment and that analysis can be used to predict performance of a newly admitted student. This research mentions an analysis of existing ML techniques in higher education for prediction of students’ success based on previous work done in the related area. We have studied a total of 40 relevant papers in which ML techniques are being proposed or implemented for students’ success prediction. A systematic analysis was done to synthesize and report the main results.
