摘要
对由Vergara等人收集的气体传感器数据集进行处理,选取手工设计特征法和主成分分析法两种不同的特征提取方法,再结合Z-score数据标准化方法构成两种特征工程方案进行实验.把两种方案处理后的数据输入人工神经网络模型中展开训练、测试.预测结果表明,主成分分析法结合Z-score标准化的特征工程方案模型准确率更优.在神经网络结构不变的情况下,选择不同的特征提取方法处理数据可以使模型预测率得到提升.
The gas sensor data set collected by Vergara et al.was processed,select two different feature extraction methods:the hand-designed feature method and the principal component analysis method,and then combined with the Z-score data normalization method to form two feature engineering schemes for experiments.The processed data of the two schemes were fed into the artificial neural network model for training and testing.The prediction results showed that the model with the principal component analysis method with Z-score normalization is more accurate.The prediction rate of the model could be improved by choosing different feature extraction methods to process the data with the same neural network structure.
作者
史雪莹
邹亚囡
吴青云
SHI Xueying;ZOU Yanan;WU Qingyun(School of Information and Control Engineering,Jilin Institute of Chemical Technology,Jilin City 132022,China;School of Science,Jilin Institute of Chemical Technology,Jinlin City 132022,China)
出处
《吉林化工学院学报》
CAS
2022年第11期1-5,共5页
Journal of Jilin Institute of Chemical Technology
基金
吉林省教育厅科研项目(JJKH20220241KJ)
关键词
数据集处理
特征提取
人工神经网络
data set processing
feature extraction
artificial neural network
