摘要
对75所高校提供的各指标原始数据,通过层次分析法确定各评价指标权重、建立模块化神经网络模型、并利用交叉验证法挑选模型,对75所高校进行资产管理绩效综合评价,并与单一神经网络模型的正确率进行比较,发现模块化神经网络模型正确率较高,且模型结果合理。
Based on the raw data of each index provided by 75 colleges and universities, determines the weight of each evaluation index by AHP, establishes a modular neural network model, and uses the cross validation method to select the model to evaluate the performance of 75 colleges and universities. Compared with the single neural network model, finds that the modular neural network model is more accurate and the model is reasonable.
出处
《现代计算机》
2018年第4期62-64,共3页
Modern Computer
关键词
层次分析
模块化神经网络
交叉验证法
综合评价
AHP
Modular Neural Network
The Cross Validation Method
Comprehensive Evaluation