Stock Market Prediction Using ARIMA, ANN and SVR

Abstract

The forecasting and estimation is the process to estimate the future price of the market stock as well as other financial commodities during the exchange. The efficacious estimation of the company’s stock price may yield fruitful results for the company in term of their increased turnover. The efficient-market hypothesis advocates that current price of the stock market be a sign of all presently accessible information and a little change in the stock market price are not based on not only the newly revealed information thus are inherently unpredictable and irregular. Others deviate and those with this viewpoint possess myriad models, methods and expertise which purportedly permit them to estimate future price information. Machine Learning methods such as Support Vector Regression (SVR), Artificial Neural Network (ANN) and other models may be thought of as mathematical function approximators. The most familiar form of ANN for stock market prediction is the feed forward network employs the backward propagation of the errors algorithm to update the network weights. The dataset for the proposed work has been collected from MSFT (Microsoft Inc) in which historical daily prices data is taken and all stock price data is kept for deliberation. The proposed work is based on the development of the stock prediction model based on SVR.