Stock Price Prediction using Mean Normalization and Neural Networks
Rohit Kriplani, Sankar Ganesh · 2015
Abstract — This paper proposes neural networks a non-linear approach to predict future trends of stock market. The model is made up of neural networks whose data are pre-processed using mean normalization technique scaled between-1 to 1. The model is created on MATLAB in the form of simulation work. The method used here is multilayer perceptron taking different number of hidden nodes into account. The Reliance stocks is considered. The sum of squares of differences between target and predicted errors is computed with 4-3-1, 4-5-1 and 4-7-1 neural network architecture. Key words: Normalization, Backpropagation algorithm, error computation I.