Hybrid usage of artificial neural network, fuzzy DANP and assurance region dea model for measuring innovation performance of country

In today's conditions where environmental changes accelerate, the competitive environment has moved to international level and countries have begun to compete with each other. One of the important competitive tools that countries have given importance in recent years and used as a strategic resource in order to have a stronger economic system than other countries and to achieve sustainable growth are R&D and innovation oriented activities. All countries invest in R&D and innovation continuously, organize awareness studies about it and develop policies that encourage innovation. In this study, a model is developed to measure the innovation performance of countries using Artificial Neural Networks (ANN), Fuzzy DEMATEL Based Analytical Network Process (FDANP) and Assurance Region Data Envelopment Analysis (DEA/AR). Taking into account the 21 innovation performance indicators studied in the Global Innovation Index reports, the relationship between these criteria is determined with YSA firstly. Then these relations are used in the FDANP method and the importance ratings of the criteria are calculated and finally, these importance scores are used as a weight constraint in the DEA/AR method to measure the innovation performances of 104 countries in 2012, 2013, 2014 and 2015. The results obtained from the study are examined comparatively with both the Global Innovation Index and the classical DEA results. It is concluded that the proposed model offers more realistic, more objective solutions than classical methods and even better distinguishes countries that have a higher performance. Keywords: Artificial Neural Networks, Fuzzy DEMATEL Based Analytic Network Process, Weighted Constrained Data Envelopment Analysis,Assurance Region, Innovation Performance