CHOICE OF METHODS AND DATA IN USING ARTIFICIAL NEURAL NETWORK ARCHITECTURE UNDER NORMALIZED AND STANDARDIZED DATA ANALYSIS OF INWARD FDI TO INDIA FROM OTHER BRICS NATIONS

Abstract

The choice of method in predicting the causation between dependent and independent variables has always been challenging for researchers. Methods' importance, ease, and robustness have eluded the community for a long. In this paper, we try to examine the use of Artificial Neural network (ANN) architecture as an alternative to the standard linear association processes and check whether the ANN architecture yields good results in the first place and what type of ANN architecture to use. Two architectures, namely Radial Bias Function (RBF) and Multi-layer Perceptron's (MLP) have been used under two different data structures, i.e., Standardized and Normalized. The results of the study show that the MLP with a standardized data set provides better results than the other methods, besides showing that the GDP of India is affected by the inflow of FDI from China and Russia and not so much from Brazil and South Africa.