Forecasting stock close price using machine learning models

Abstract

Forecasting is about determining which variables contribute to the prediction of other variables. The stock forecasting problem is difficult due to its unpredictable and non-linear nature. Prediction and evaluation are two of the most challenging tasks. A machine learning (ML) classifier was used to address prediction issues in this paper. Linear Regression (LR) was used to develop the new Stock Close Price Prediction (SCPP) algorithm. The model was built using Yahoo Finance's data of different stocks using LR, Bernoulli Naïve Bayes (BNB), and Support Vector Machine (SVM). Among these, the model forecasts the stock close price effectively with LR. In this, the Apple stock’s close price was predicted with the features: Low, Open, High, Volume, and Close. From the results, LR outperforms the SVM and BNB in close price prediction with an accuracy of 99.97 %.