A Hybrid Prediction Model of Kernel Principal Component Analysis, Support Vector Regression and Teaching Learning Based Optimization Techniques

Abstract

Forecasting of stock market is considered as one of the most decisive and critical tasks for the data scientists in financial domain. Stock market is one of exciting and demanding monetary activities for individual investors, and financial analysts. The stock market is an inter-connected important economic international business. Prediction of stock price has become a crucial issue for stock investors and brokers. The stock market is able to influence the day to day life of the common people. The stock price is based on the state of market stability. As the dormant high noises in the data impair the performance, reducing the noise would be competent while constructing the forecasting model. To achieve this task, integration of kernel principal component analysis, support vector machine with teaching learning based optimization algorithm is proposed in this research work. Kernel principal component analysis is able to remove the unnecessary and unrelated factors, and reduces the dimension of input variables and time complexity. The feasibility and efficiency of this proposed hybrid model has been applied to forecast the daily open prices of stock index of a leading Company. The performance of the proposed approach is evaluated with 3543 daily transactional (13th December 2001 to 4th December 2020) stocks price data from Bombay Stock Exchange (BSE). Empirical results show that the proposed model enhances the performance of the prediction model and can be used for taking better decision and more accurate predictions for financial investors.