Evaluation of the trade competitiveness of colombian industrial sector through the coefficient of export opening

Abstract

In the present investigation an analysis of Colombia's trade competitiveness from opening export ratio registered by the National Planning Department in the period 2001 to 2012. For the investigation was an application of a probabilistic neural network is made by the statgraphics centurion software, which required designing a neural network with 10 input neurons, one for each input variable; 10 cases of training associated with training periods 2001 to 2010 and three output neurons associated with competitive ranking levels defined by expert judgment network. Samples for the periods 2011 and 2012 were reserved to validate qualifying ability of the neural network. Overall we noted the qualifying network capacity, reaching 95.23% probability of correct classification. Once the neural network was trained, the reserved test samples could be successfully classified.