Forecasting research trends using population dynamics model with Burgers' type interaction

The increasing costs of research and the decreasing lifetime of products and processes make the decisions on allocation of R&D funds strategically important. Therefore, ability to predict research trends is crucial in minimizing risks of R&D expenditure planning. The purpose of this paper is to propose a model for efficient prediction of research trends in a chosen branch of science. The approach is based on population dynamics with Burgers' type global interaction and selective neighborhood. The model is estimated based on a training set and an out-of-sample forecast is performed. The research trends of filtration processes were analysed in this paper. The simulation results show that the model is able to predict the trends with a considerable accuracy and should, therefore, be tested on a wider range of research fields.