The Machine Learning has been used in educational area necessitated to handle several types of problems such as: to handle the drop out problems/cases, to improve the students’ retention cases, knowing in advance at risk students, to predict and analysis the students’ performance. Recently, lot of changes have occurred in education sector/system, such as school/university were temporary closed, offline education work moved towards an online education, school/university have reopened, bringing out major changes in the behavior of students which directly or indirectly affects the performance of students. Compatibility of this study to existing study for obtaining best predictive accuracy value model with significant datasets. For predictive analysis the performance of student into three categories such as excellent , average and poor with significant datasets, consequently upon reopening of schools, the aim/objective of this study for considering the selection between 1501 to 9000 range of datasets by determining the range on average bases somewhere on the point neither more nor less number of previous researchers and also identifying the exiting the best machine learning algorithms whose accuracy value may be above 90%.From 2019 to 2021 MLP (Multi-layer Perceptron), RF (Random Forest), QDA (Quadratic Discriminant Analysis), LGBM (Gradient Boosting), Support Vector Machine, Linear Regression, BiLSTM (Bidirectional Long Short-Term Memory) algorithms and to provide higher accuracy value that was greater than 90%. After the analysis of previous research work there were seven algorithms whose accuracy value above than the 90% and also the modest range of datasets (that was greater than 1500 and less than equal to 9000(>1500&<= 9000)) was considered by neither more nor less previous researchers (4 previous researchers) in their studies.

Machine Learning, Performance of the Students, Evaluation Matrix, Predictive Analytics, Education System.