Stock market prediction using machine learning algorithms

Abstract

The purpose of this work is to predict the stock price fluctuation and find the best usable algorithm for predicting stock price by comparing the outcomes of various algorithms of machine learning considering various factors. The algorithms we are proposing to use are linear regression, logical regression, LSTM, random forest algorithm, SVM and naive Bayes' algorithm. We are aiming to apply various algorithms and predict the one with best outcome. The factor which we have used as an attribute in our model is news that we assume will affect the price of the stock market. We have taken the top 25 news of the day and each news will be evaluated as if it’s positive or negative and then the positive new is assigned with +1 value and so as negative with-1. For the particular day the sum of the news values are taken if the sum is positive then the predicted price will increase as compare to previous price but if the sum is negative the value should decrease base on these facts evaluation is done between predicted and occurred value and so the algorithms are used to generate the prediction and hence used to calculate the accuracy provided by the algorithm .Using news as the factor may help us in the more chance of increase in the detecting the fluctuation in the values as the news is one of the greatest factor effecting the change in stock prize as news contain the brief every possible event happened in the previous day and also contain about the company that is their release of product, status, bonds, funds, investments.