Application of Artificial Neural Networks in Predicting Macro Indicators of Science and Technology

The researches carried out on R & D assessment and the relationship between production of science and technology at the macro level of countries were limited due to the high volume of information and rapid rate of changes. This paper presents application of data mining algorithms for modelling technology indicator of countries based on their science production. Furthermore, the impact of any scientometrics indicator on the technology indicator was separately determined using sensitivity analysis of neural networks. In the research, the number of patents registered by countries in the World Intellectual Property Organization (WIPO) was considered as technology indicator and data collected from the SCImago Journal & Country Rank (SJCR) was considered as indicators of science production. Data were ranged from 2001 to 2015. The artificial neural networks and regression were used to model the data. According to validation results, the artificial neural networks had more accuracy for modelling the science and technology indicators. The results of sensitivity analysis showed that the H-index parameter was the most important indicator for predicting the technology indicator. Predicting the technology is the first stage for Planning at the macro level and the purpose of that is making better decisions and programming for future.