A Research On Prediction Of Academic Record Based On Neural Network And Cover Algorithm

With the development of information technology, it is necessary for teachers to explore new ways of learning, teaching methods and to construct a digital campus. It is important for every teacher to use information technology, reasonably, appropriately and effectively, in order to solve some problems in the conventional teaching. The analysis of student's grades such as average points, passing rates, excellence rates, and the highest score and so on is often done through the computer. However the analysis fails to show the relationships and rules that exist between them, and predict student's grades and student developmental tendency in the future. As a school, it is an important proof for the school to evaluate the quality of teaching based upon students'grades analyzing.If we can figure out the greater factors impact on the performance that have students from the existing grades and other relevant information, and make certain predictions, it will be of a great help to improve the way we cultivate students and strengthen the quality of teaching in schools.The traditional method of data mining in education and teaching is called decision tree, in this way we can figure out the factors that affect students grades and the relationship between the factors, but this method may be more suitable for dissertation deals with the analysis of the students'grades in institutions of higher learning.This paper deals with experiment-based cross-covering algorithm aiming to find the more important factors that influence middle school students' learning. Cross-covering algorithm is a field covering algorithm, which obtained from the geometric meaning of MP neuron model. In a sense, it takes into accounts the optimization of network structure, causing a smaller neural network and this method is practical. It solves the problem of how to design a feedforword network, and this problem has not been well solved for many years after all.At the same time, after comparing the experimental results of cross-covering algorithm and standard BP algorithm, we make a conclusion that cross-covering algorithm is faster and more accurate than standard BP algorithm. It can be seen from the dissertation, the neural network algorithm will be applied to research and analysis of student's grades and comprehensive quality in order to improve teaching quality. This paper carried out researches in to the following areas:1. We made a study of current situation of domestic and foreign research about data mining.2. We grasped the basic concepts and methods of data mining, and the current research and applications of data mining. We study neural network algorithm to figure out which algorithm is better for solving the problem.3. We collected grades relevant information and examination results, and analyzed their grades database, in order to find out the demerits of current performance evaluation. By analyzing the data of teaching assessment, we provide decision supporting information for teaching departments.4. Based on database of students'achievement, we carried out the data collection, clean-up, making it suitable for data mining. Then we used cross-covering algorithm to analyze students'grades, and predict results.5. Finally, we illustrate the implementation of data mining technology and application of cross-covering algorithm in management system of secondary education and instruction.