Repair and prediction of stock market data with effective neura network models

Abstract

modeling is an open problem ever. In stock price prediction, simultaneous newlineachievement of higher accuracy and the fastest prediction becomes a challenging newlineproblem due to the hidden information found in raw data. Various prediction newlinemodels based on machine learning algorithms have been proposed in the newlineliterature. The performance of such learning algorithms heavily depends on the newlinequality of the data as well as optimal learning parameters. Among the newlineconventional prediction methods, the use of neural network has greatest research newlineinterest because of their advantages of self-organizing, distributed processing newlineand self-learning behaviors. In this work, dynamic nature of the data is mainly newlinefocused. In conventional models the retraining has to be carried out for two newlinecases: the data used for training has higher noise and outliers or model trained newlinewithout preprocessing; the learned data has to update dynamically for recent newlinechanges. In this sense, it is proposed to create a self-repairing dynamic model newlinecalled Repairing Artificial Neural Network (RANN) that correct such errors newlineeffectively. newlineThe repairing includes adjusting the prediction model from noise, newlineoutliers, removing a data sample, and adjusting an attribute value. Hence, the newlinetotal reconstruction of the prediction model could be avoided while saving newlinetraining time. The proposed model is validated with five different real-time stock newlinemarket data and the results are quantified to analyze its performance. The newlineperformance of the proposed model is validated with five standard stock market newlinedata sets such as Nifty 50, Nifty Bank, Nifty Pharma, BSE IT, and BSE Oil and newlineGas. Data of five years that are collected for each dataset, and the stock price newlineforecasting performance are measured with three error rates and three prediction newlineaccuracy measures. The RANN model is compared with the existing five newlinedifferent neural network models. The investigated results have shown that the newlineRANN model is achieving lower error rates and higher prediction accuracy newlinewhile adopting dynamic changes.