Contents I. Introduction
The concept of the neural network model is based on the brain's anatomy; when the human brain received biological signals for projecting accurate perception, the intermediate model simulates its judgment. The input signals calculated its weight and passed another hidden layer for making accurate projections to the brain. The classification techniques are carried out when there is a new category will be predicted after having various attributes of input data of student grades. Al-though student's grade prediction is based on large parameters and on study time management. Worldwide there was a large number of parameters were being applied to measure current student grades. However, none model accurately predicts future grades. This research tries to predict student grades using backgrounds and current student grades for the final GPA of an engineering student. Similarly, the type of grade prediction research carried out at The Islamic University of Bahawalpur concluded that 16 students who scored 80 marks in internal assignments were failed in the final. However, this study covers master in education students [1]. Similarly, [12] applies the ID3 algorithm using a sample of 50 students to identify various attributes to perform best scores. [2] Used linear regression on a sample of 300 students from affiliated colleges of Panjab University. The study [8] surveyed 1847 students in Nigerian Universities concluded that the internal consistency reliability of grade was 0.86 used heterogeneous classes in university students used the Naive Bayes algorithm to predict student performance using data repository. Which used 220 students' samples, used175 records as training, and 45 records were test achieve an 86.66% model accuracy model using MATLAB. Similarly, [7] studied farmers' crop prediction in Maharashtra soil quality, soil nutrients for crop production using Multiple Linear Regression (MLR). The decision tree algorithm was used to perform an analysis of 362 datasets of crops. The training dataset for classification of organic, inorganic soil. The backpropagation concept uses a hidden layer for better prediction of soil properties. The Support Vector Machine (SVM) and Relevance Vector Machine (RVM) were chosen as effective predictions for smart wireless sensing of meteorologi-cal data concludes an error rate of 15% and 95% accuracy [11]. Likewise, [?]the ice forecast for the green with considering different methodological factors temperatures ice assurance of natural products and vegetables of machine learning producers as a rule utilized in relapse and classification issues counting artificial neural network used Bayes classifiers, and k-Nearest Neighbors. The proposed framework with the precision of 98.9% in anticipating cancer repeat over three cancer datasets. It is one of the deadliest infections all-inclusive and is capable of around 13% of all passing around the world [9]. The cancer incidence rate is developing at a disturbing rate within the world Naive Bayes, which is in a general sense built on the solid freedom suspicion, could be an exceptionally well-known classifier in machine learning due to its productivity [10]. At the Government College of Horticulture, Abeokuta, Nigeria, utilize ANN for foreseeing students' grades based on the past datasets comprised of 60 understudies within data science, who have completed four scholarly sessions from the college using MATLAB. ANN model accurately anticipates students' last grades with 91.7% precision for CGPA expectation. In this work, different statistic highlights (age, sexual orientation, and race), as well as college entrance examination, were utilized to prepare a neural network to foresee the ultimate understudy review of understudies in e-learning courses uti-lizing numerous feed-forward neural systems and multiple-choice test information managed to the understudies of National Specialized College, Athens, Greece as input. Building course. Similarly, [5]researched massive online courses of pre-student performance prediction found 60 percent accuracy of average performance using neural networks. The predictive learning networks that predict learning outcomes in different points ultimately informed poor student attention from an instructor and caused drop-out in early attention. However, the greatest challenges of grade prediction are that many students may choose a single alternative answer in multiple type questions [6]. Moreover, the learning process may depend on many personalized motivations, which largely increased due to online materials increasing 61% more RMSE and, on average more than 10 percent importance in grade performance. Similarly, a study conducted on medical students with 40 sample studies concluded a positive correlation between attendance and marks in all the internal theory and practical examinations. However, this research was conducted on under, high school grade, parents' educations level, plus2 grade, and three major subjects (Physics, Math's, Chemistry) the previous score were backgrounded independent variables and internal assignments marks, of ongoing semesters, were passed for student grade using various types neural model for predicting student university grades [3].
