摘要
将模糊信息粒化、支持向量回归机与数据包络分析相结合,可以预测未来时间段某一地区科技创新的效率。以浙江省为例,选取1990~2018年科技创新效率的投入产出指标作为研究对象,利用SVR分别对用模糊信息粒化处理后的数据和原数据进行回归预测;为进一步增强预测的准确性,采用K折交叉验证法和遗传算法分别对各组数据进行参数优化,选择拟合效果更好的参数对SVR进行训练并建立最优回归模型;最后,利用DEA计算预测数据的效率。实验结果表明FIG-SVR模型比单一的SVR模型能够提高预测精度及范围,具有更好的适用性。
By combining fuzzy information granulation(FIG),supporting vector regression(SVR)and data envelopment analysis(DEA),the efficiency of technological innovation in a certain region can be predicted in the future.Take Zhejiang Province as an example,selecting the input-output indicators of scientific and technological innovation efficiency from 1990 to 2018 as the research object,and using SVR to perform regression prediction on the data granulated with fuzzy information and the original data;in order to further enhance the accuracy of the prediction.Using K-fold cross-validation method and genetic algorithm to optimize the parameters of each group of data,selecting the better fitting parameters to train the SVR and establishing the optimal regression model;finally,using DEA to calculate the efficiency of the predicted data.The experimental results show that the FIG-SVR model can improve the prediction accuracy and scope than a single SVR model,and has better applicability.
作者
李争艳
王向前
LI Zheng-yan;WANG Xiang-qian(School of Economics and Management,Anhui University of Science and Technology,Huainan 232001,China)
出处
《河南科技大学学报（社会科学版）》
2021年第5期31-38,共8页
Journal of Henan University of Science & Technology(Social science)
基金
国家自然科学基金项目(51874003)
安徽理工大学研究生创新基金项目(2020CX2091)。
关键词
模糊信息粒化
支持向量回归机
数据包络分析
参数优化
效率预测
fuzzy information granulation
support vector regression machine
data envelope analysis
parameter optimization
efficiency prediction
