The current living environment around us is driven by making investment and having good return on investment. People trade shares of public listed companies as buyers or sellers where they have rewarding outcome. But the nature of the existing Stock Market is very volatile and unpredictive which imposes greater risk of decision making. The computing sector is trying its best to construct a predictor model which can precisely forecast Stock Market prices under all circumstances. Main fluctuating factors in stock market prices are numerous. It can be phycological (like capitalist sentiment) or political and business related (like budgetary news) and environmental disasters (like natural calamities) and other ongoing events. All these factors contribute to increased complexity of the predictive model. Consistently experiments are undertaken and executed with a number of methods involving Machine Learning, Deep Learnings and many Time-series approaches to build an explicit predictive model. In this project, time-series model, known as ARIMA (Autoregressive integrated moving average) model and deep learning model, known as LSTM (long-short term memory) model are executed. These Deep-Learning models are implemented because they can identify paradigms and insights remarkably. The aim of the experiment is to see whether regressive or time-series model gives higher precision and accurate forecasts by comparing the prediction values and find the best fit.

Keywords: Machine Learning, Deep Learning, Data Science, Autoregressive integrated moving average (ARIMA), LSTM (long-short term memory), Stock Market Prediction
View Full Paper