基于模糊C均值面板数据聚类的中国省域专利产出分析

本文首先对三维空间中面板数据的曲面几何特征进行了描述,并从"绝对数量"、"增长速度"、"几何相似性"和"空间绝对距离"等几个方面对曲面相似性指标进行了定义和构建,对模糊C均值聚类方法进行了优化.在此基础上,对2000-2010年中国31个省市专利产出活动的类型特征及其地域分异规律进行了探索.实证研究结果表明聚类效果良好,中国专利产出无论数量、绩效还是增速在总体上均呈上升趋势并存在空间异质性和自相关性.创新能力较强的省区虽集中在东部,但正在向中西部地区扩散.同时,创新总体上仍主要来自于政府推动尤其是研发资金投入,并且研发资源投入总量以及研发人员可支配资金的区域配置极不均衡.此外,财政拨款对于东、西部地区创新效率的作用也存在很大差异.因此,政府可根据区域研发能力和资源现状的不同,制定合理的区域科技发展战略与相关政策工具,从而挖掘区域创新动力,提升区域专利创新能力.

Chinese provincial patent product analysis based on fuzzy C-mean clustering method for panel data

This paper described the geometric structure of panel data's curved surface in 3D space, and established the synthesized similarity index for the curved surface through several aspects including the absolute magnitude, growth rate, geometric similarity and spatial absolute distance. The traditional fuzzy C-means clustering method was optimized through replacing the distance with the similarity index. On this basis, this paper explored the patent output's type characteristic and the rule of its territorial differentiation in China's 31 provinces (municipalities) during the years of 2000-2010. The empirical results showed that the clustering effect was good, and Chinese patent output showed an upward trend and spatial heterogeneity and autocorrelation generally at the same time no matter in the number, the performance or the growth rate. The innovation capability of more and more regions in central and western China were getting enhanced, although regions with higher innovation ability still concentrated in the eastern China. In addition, the boost of innovation mainly came from governments' promotion especially the R&D funding investment, however, the allocation of R&D resources and the personnel disposable funds were both extremely unbalanced among regions. Besides, there were distinct differences in the innovation efficiency between eastern and western regions for the financial allocation from government. The government should formulate a reasonable regional scientific and technological development strategy and sets of policy tools according to the regional R&D ability and resource endowment, so as to exploit the regional motivation power of innovation and to improve regional patent innovation capability.