关键词: 高考成绩预测, 数据挖掘, 神经网络, 多元特征

Abstract: College entrance examination grades are the focus that attracted the attention of major education departments today,scientific grade analysis can help teachers and students in the school to arrange and adjust the study plans reasonably to improve the grades of candidates.Traditional methods use the knowledge of statistics and data mining to discover the implicit connections between grades,the accuracy of which is limited faced with a variety of nonlinear data.By combining the candidates’ features of short-term and long-term,the key factors that affect the grades of the college entrance examination were fully tapped,and a novel Multiple Features-aware Neural Network (MFNN) model was proposed to achieve the grades of the college entrance examination prediction.To verify the effectiveness of MFNN,the experiments on the real data set provided by Hefei Education Bureau was conducted.The data set included three high school quality test grades and college entrance examination grades of 10138 science and engineering candidates and 4874 literature and history candidates in 2015.Experimental results showed that the proposed method was superior to other comparison methods.

Key words: college entrance examination prediction, data mining, neural network, multiple features