Prediction of students educational academic recognition using neural networks

The swift development of information as well as communication technology, along with the proliferation of mobile devices, has revolutionized education. The study, which is reported in this article, compared various machine learning techniques for forecasting academic performance. A compromise between accuracy and interpretability was reached through our decision-making process, and this balance was ideal for identifying trends in high school students’ academic performance. We employ six supervised learning algorithms: Random Forest, Logistic Regression, Gradient Boosting, Light Gradient Boosting Machine, AdaBoost, and K-nearest Neighbors to identify patterns. Gradient boosting has a greater accuracy of roughly 96.77% in the random old out approach, while LSTM has the lowest accuracy at 84%, as is plainly demonstrated. In shuffle 5-fold cross validation CNN has accuracy of 95.10%. and Gradient boosting has least accuracy of 80.12%.