Using SOM for Analyzing and Interpreting the Clustering and Characteristics of Nascent Entrepreneur

碩士 === 長榮大學 === 經營管理研究所 === 98 === Although researches on entrepreneurship across countries have been a hit in the last decade, the factors influencing entrepreneurs are various yet the analysis approaches are still limited. Previous studies had helped us to define numerous factors affecting nascent...

Full description

Bibliographic Details
Main Authors: Chung-Wei Lin, 林仲緯
Other Authors: Li-Ming Chuang
Format: Others
Language:en_US
Published: 2010
Online Access:http://ndltd.ncl.edu.tw/handle/85nz52
Description
Summary:碩士 === 長榮大學 === 經營管理研究所 === 98 === Although researches on entrepreneurship across countries have been a hit in the last decade, the factors influencing entrepreneurs are various yet the analysis approaches are still limited. Previous studies had helped us to define numerous factors affecting nascent entrepreneurs, it can be argued that our empirically based knowledge about nascent entrepreneurship is still limited (Davidsson and Honig, 2003). Therefore, Davidsson and Honig (2003) integrated and demonstrated two major dimensions of factors affecting nascent entrepreneurs, Human Capital and Social Capital. Besides, traditional statistical methods within mere correlation analyses on entrepreneurship researches are insufficient to analysis long period preceding data. Prior research indicated some research limitation of entrepreneurial research. First, Arenius and Minniti (2005) indicated that their data did not allow them to establish the causal direction between conceptual variables and entrepreneurial behavior. Moreover, they indicated that treatment of country effects was limited to the introduction of dummy variables. Therefore they gave a future research direction that those who follow their study could lead to an unambiguous understanding of how perceptions influence entrepreneurial behavior by following their explorative investigation and use the data including macroeconomic factors, such as technological sophistication and culture. Second, Engelen et al (2009) indicated that in the cross-cultural studies, the underlying mechanisms in terms of the causes and effects in differences are still disturbingly unclear, future studies need to build more complex research models that go beyond mere correlation analyses of certain phenomena and national culture. Furthermore, Sternberg and Wennekers (2005) indicated that in an analysis focusing on the effects of entrepreneurship, data on entrepreneurship should be available for sufficiently long period preceding the measurement of dependent variables therefore time series data on entrepreneurial activity for a large number of countries or longitudinal data on individuals may shed more light on the factors determining entrepreneurship. In this research, we use data from Global Entrepreneurship Monitor (GEM) to examine 14 factors on 10 countries by employing Kohonen Self-Organizing Maps (SOM) as a mean to identify the patterns exist between selected countries and years (through year 2001 to 2006). The Global Entrepreneurship Monitor is an ongoing multinational project trying to detect why entrepreneurial activities vary across countries, and how entrepreneurial activities affect economic growth. We attempt to use this harmonized, international comparable data on entrepreneurial activities as our data source. And by the use of SOM, a neural network algorithm that is applied in pattern recognition, image analysis, process monitoring and fault diagnosis, to solve the limitation of statistical methods. Our result identified 4 clusters (courageous, experienced, conservative and compensative) and announced some features associated with each. This result illustrates some specific patterns of entrepreneurs and shows how countries shift overtime. This research provides some implications: ie., countries in compensative cluster should enforce their government policies that strengthen confidence of entrepreneurs, therefore they may conquer the anxiety of being failure; entrepreneurs in courageous cluster should develop their social networks, which may help them get better supports from others.