Stock price prediction using time series data

Abstract

Researchers has taken a lot of years to make algorithms fast and accurate enough to make stock price predictions accurately. Investors are looking for smarter techniques to forecast stock prices for investments and this has made this topic one of the most worked out researches in data science eld. One of the trendy ways of forecasting is time series analysis. In this thesis, I have compared recent 3 most common time series forecasting algorithms that are- Autoregressive Integrated Moving Average, Facebook prophet and Long Short Term Memory, using company data (LMT and NOC) from yahoo nance. Firstly, I used K-Means clustering to choose a cluster with least number of companies and then used processed data to compare the accuracy of the algorithms.