| 基于IG-DNN混合决策算法的糖尿病预测研究 |
| |
| 引用本文： | 卢春城,黄理灿,刘靖雯.基于IG-DNN混合决策算法的糖尿病预测研究[J].教育技术导刊,2019,18(8):21-25. |
| |
| 作者姓名： | 卢春城 黄理灿 刘靖雯 |
| |
| 作者单位： | 浙江理工大学 信息学院，浙江 杭州 310018 |
| |
| 摘 要： | 糖尿病患者人数众多，对人的健康危害极大，尽早预测是否患有糖尿病是降低糖尿病死亡率的关键。基于IG-DNN混合决策算法进行糖尿病预测模型研究，其中糖尿病数据集来源于UCI机器学习库—PIDD。PIDD包括768个记录，每条记录包含8个属性。首先应用信息增益方法（IG）将属性减少到5个，然后将其应用于DNN作为输入。该方法分类准确度达到88.3%，效果优于之前的大部分研究成果。
|
| 关 键 词： | 糖尿病预测模型 PIDD 信息增益（IG） 深度神经网络（DNN） |
| 收稿时间： | 2018-12-11 |
Research on Diabetes Hybrid Decision Algorithm Based on IG-DNN |
| |
| LU Chun-cheng,HUANG Li-can,LIU Jing-wen.Research on Diabetes Hybrid Decision Algorithm Based on IG-DNN[J].Introduction of Educational Technology,2019,18(8):21-25. |
| |
| Authors: | LU Chun-cheng HUANG Li-can LIU Jing-wen |
| |
| Institution: | School of Information Science and Technology， Zhejiang Sci-tech University， Hangzhou 310018， China |
| |
| Abstract: | A large number of patients suffer from diabetes which is extremely harmful to people’s health， and early prediction of diabetes is the key to reducing diabetes mortality. Machine learning algorithms are often used to build diabetes prediction models. In this paper， a hybrid decision algorithm based on IG-DNN is proposed. The diabetes dataset was derived from the UCI machine learning library， PIDD. The PIDD consists of 768 records， each of which contains 8 attributes. The proposed new method first applies the information gain method （IG） to reduce the attribute to 5 and then applies it to the DNN as input. The classification accuracy of the proposed new method is 88.3%， which is better than most previous research results. |
| |
| Keywords: | diabetes prediction model PIDD information gain deep neural network |
|
| 点击此处可从《教育技术导刊》浏览原始摘要信息 |
| 点击此处可从《教育技术导刊》下载免费的PDF全文 |
