Cat Swarm Optimization Algorithm Tuned Multilayer Perceptron for Stock Price Prediction

Abstract

Due to the non-linear and dynamic nature of stock data, prediction is one of the hard tasks in the financial market. Now, soft and bio-inspired computing algorithms have been used to forecast the stock price. This article assessed the efficiency of the hybrid prediction model using multi-layer perception (MLP) and cat swarm optimization (CSO) algorithm. CSO algorithm is a kind of bio-inspired algorithm motivated by the behavior traits of cats. CSO is employed to find appropriate value of MLP parameters. Technical indicators calculated from historical data are used as input variables to the proposed model. The performance of the model is validated by using historical data not used for training. The prediction efficiency of the model is evaluated in terms of MSE, MAPE, RMSE, and MAE. The results of the model are compared with other models optimized by various bio-inspired algorithms explored in the literature to prove its efficiency. The empirical findings proved that the proposed CSO-MLP prediction model provides best performance when compared to other models taken for analysis.