Prediction and Clustering Techniques used in the Development of Stock Forecasting Model

Abstract

With the advancement in science and technology more and more real time data is being accumulated in the digital repositories. One such highly accumulating data is stock market data. Prediction of stock market data and its analysis is a challenging task as it is a highly sophisticated time series data vulnerable to sudden changes. The data and its relevance in the real world has attracted the interest of many researchers. Research literature provides many contributions by eminent researchers for analyzing and developing models for stock data. In this research paper an effort has been made present two things. One, building a stock prediction model using artificial neural network using different learning functions and we found that the different learning functions produces different results. In our experiment we have achieved a highest accuracy of 94.55%. The second thing we are trying to do in this article is to provide a detailed information on various techniques used for stock data prediction. More than 50 articles have been studied to present the significant contribution of the researchers. The articles are categorized into two main sections namely prediction methodologies used for stock data analysis and clustering methodologies used for stock data analysis. The sections, prediction methodologies used for analysis and clustering methodologies used for analysis are further explored into eight and four sub categories respectively depending upon the methodologies used. After presenting the detailed analysis, we have also highlighted the research gaps existing in the methodologies discussed in this paper.