Stock market price variation prediction using mathematical models and machine learning techniques

Abstract

The stock market is an area where the stock prices are not constant; however, they are dynamic, nonlinear and volatile in nature. The prediction of the stock price is a challenging task with the factors such as the financial performance of the organization, politics, unexpected natural and man-made causes, and worldwide economic conditions. In concern to these issues, several analytic techniques are being developed by researchers, financial analysts, and data scientists for exploring the nature of stock market trends. Most of the researchers from the past followed two types of stock price prediction techniques such as (i) Technical Analysis and (ii) textual analysis. The former follows analyzing the price direction to predict future prices and the latter follows analyzing the financial news and earning reports to predict the price. The former is an effective one to solve the issues of stock price prediction. The works proposed here follow three novel objectives to predict stock prices. newlineIn phase I a novel approach is known as the Heuristic optimization-based hybrid approach for the pre-processing of data collected from the stocks. The GANN square technique for the prediction of stock price exactly. The data pre-processing involves four stages (i) ignoring the data with missing values of features, (ii) data imputation initialization, (iii) detection and deletion of outliers, and Hybrid Archimedes-based Salp Swarm Algorithm based future selection. Thus the stated approach effectively removes the unwanted noises and features from the dataset. this made it easier for the GANN square technique the prediction of stock prices effectively. newline