ComparativeAnalysisofARIMAandLSTMM achine Learning Algorithm for Stock PricePrediction

Abstract

Stocksofcompaniesheavilyinfluencethefinancial markets around the world. These companies help tocontributeandimprovetheoverallGDPofaneconomy.Hence, the importance of having a grip on the stock market forventurecapitalistsandcompaniesisinevitablefortheirfinancial benefit and growth. It is crucial to predict the stockprice to stay at the forefront of the financial world. None of theexistingmachinelearningtechniquescanprovideaperfectpredi ction of the stock prices due to the unpredictable identityof the stock market. The stock price prediction employing twomachinelearningalgorithms,LongShort-TermMemory(LSTM)andAutoregressivelntegratedMovingAve rage(ARIMA), willbediscussedindepthinthisstudy. Theaccuracy achieved by these two algorithms was compared. Inour comparison, we found out that, generally, LSTM had ahigheraccuracyrateinthestockpriceprediction.ARIMAprovide dbetterperformancewithasmalldatatimeframe, while LSTM had better performance in predicting stock pricewhenthedatatimeframeusedwaslarge.