Financial market prediction under deep learning framework using auto encoder and kernel extreme learning machine

Abstract

The technical indicators are highly uncertain therefore possess greater influence on the stock market prediction. Among different techniques developed for effective prediction of the financial market the AI techniques show better prediction efficiency. In this paper, a hybrid model combined with auto encoder (AE) and kernel extreme learning machine (KELM) is proposed for further improvement in the quality of financial market prediction. This study mainly emphasizes on a precise prediction of the financial market, the main motive behind stock price prediction is minimizing the substantial losses faced by investors, and analysing the profitability with the help of buying and selling amount. The prime advantage of the proposed technique over the conventional SAE is robust prediction of different financial market with reduction in error. To authenticate the performance of the proposed deep learning (DL) technique (KELM-AE), high-frequency data of different financial market like Yes Bank, SBI, ASHR, and DJI are taken into consideration and the performance of the proposed technique is investigated in MATLAB based simulation in accordance with MAPE (Mean Absolute Percentage Error), MAE (Mean Absolute Error) and RMSE (Root Mean Square Error). The application of SAE is new in the field of predicting different bank data. The validation of the model is performed by comparing it with other traditional methods based on different performance indexes. The simulation result indicates that the proposed DL based technique (KELM-AE) outperforms other models with a MAPE value of less than 2%for future prediction, irrespective of the financial market. For example the MAPE value for KELM-AE is observed to be 1.074 %, 0.888%, 1.021% for YES, SBI and BOI respectively which is much lower as compared to other model like ELM that shows a MAPE value of 1.714%, 1.473% and 1.550% for the above mentioned bank.