基于DNN神经网络的研究生生源预测方法研究

Research on the Method of Graduate Student Source Prediction Based on DNN Neural Network

高校研究生的生源质量是提高研究生培养水平的基础,获得准确的生源质量信息可以为招生宣传提供科学有效的背景支撑与技术保障。本文以吉林大学某学院近五年招生生源类型为样本,建立高校研究生生源质量评价指标体系,基于DNN神经网络拟合高校生源质量变化特点,在谷歌研发的Tensorflow框架下建立高校研究生生源质量评价模型;最后,采用训练完备数据库对下一年研究生生源质量进行预测分析,指导招生宣传工作的高效执行。本文所提预测方法较好地解决了研究生生源质量难以保障研究生培养水平的问题,对提高高校研究生整体培养水平以及相应学科建设的发展具有重要意义。

The quality of graduate students is the basis and guarantee to improve the quality of graduate training. Obtaining accurate quality information of graduate students can provide scientific and effective technical support for enrollment promotion. In this paper, the quality evaluation index system of graduate students in colleges and universities is established based on the types of enrollment students in a college of Jilin University in the past five years;then DNN network is introduced to fit the characteristics of the quality change of graduate students in colleges and universities, and the quality evaluation model of graduate students in colleges and universities is established under the Tensorflow framework developed by Google;finally, the data of seven years of training is used to predict the quality of graduate students in the next year. The prediction method proposed in this paper can better solve the problem that the quality of graduate students is difficult to guarantee the level of graduate education, and has certain guiding significance for the follow-up work.