基于神经网络的高校贫困生辅助认定模型研究

Research of Auxiliary Identification Model of Poverty CollegeStudents Based on NNs

“精准资助”是现阶段我国贫困生资助等教育扶贫工作的新任务,贫困生认定工作作为贫困生资助的首要环节,现行的传统流程中存在着“假贫困”、认定标准主观性强等问题。将数据挖掘应用于贫困生辅助认定,基于学生消费行为习惯、学习情况和家庭情况等相关数据,对智慧校园长期积累的数据产物进行数据采样和建模,形成贫困生特征样本数据集,利用TensorFlow对全连接神经网络进行模型训练,根据模型产生期望输出,得到贫困生辅助认定模型。随机抽取输出的测试集数据对比已有贫困生数据进行精度测试,测试准确率较高。整个模型训练过程包括数据采样、数据建模、模型训练和模型评价等过程,将其应用于贫困生辅助认定,为传统主观的贫困生认定提供了更为精准、科学、客观的决策支撑。

Precise funding is a new task of educational poverty alleviation efforts such as financial assistance for poverty students at the present stage.The identification of poverty students is the primary link of financial assistance for poverty students,while some problems exist in the current traditional process,such as“false poverty”and subjective identification criteria.Data mining is used in the identification of poverty students,and the identification is based on the students'relevant data about the consumption behavior,study situation and family status.The data products accumulated in smart campus for a long time are sampled and modeled to form the characteristic sample data set of poverty students.TensorFlow is used for model training on the fully connected neural network,the expected output is generated according to the model,and the auxiliary identification model for poverty students is obtained.The output test set data are randomly selected and compared with the existing poverty students'data for accuracy test,and the accuracy of the test is high.The whole model training process is composed of data sampling,data modeling,model training and model estimating.The model is applied in auxiliary identification of poverty students,which provides more accurate,scientific and objective decision support for traditional identification of poverty students.