Comparison of five computational intelligence algorithms to predict entrepreneurial profile from Universidad Iberoamericana León students

Abstract

Nowadays entrepreneurship is considered one of the most important developing factors in emerging economies. Therefore, it is very important to generate strategies that increase the number of entrepreneurs will graduate from universities. The present study aims to compare five computational intelligence algorithms to predict the entrepreneurial profile of students from Iberoamericana University in Leon Guanajuato. For this study 22 variables were analyzed, using statistical techniques and artificial intelligence. The variables are grouped by: family support enterprise processes, internal locus, innovation and risk-taking. The total of student surveyed were 236. Of this, 125 were from the Business Faculty, 111 students from Basic Science and Social Science Faculty. We select this profile of students because students from Business Faculty have a profile with entrepreneurial features, and students from Basic Science and Social Science faculty have a distant profile of an entrepreneur. Five classification algorithms were performed to make a comparison: Logistic, Multilayer Perceptron, Simple Logistic, J48 and Linear Discriminant Analysis.