Review Paper on Stock Price Prediction using Supervised Machine Learning Technique

Abstract

Stock market prediction has become a significant research domain due to its potential to support informed investment decisions, risk management, and financial planning. However, predicting stock prices remains challenging because of the dynamic, nonlinear, and highly volatile nature of financial markets. Supervised Machine Learning (ML) techniques have emerged as powerful tools for modeling complex price patterns by learning from historical, labeled stock data. This review paper provides a comprehensive analysis of various supervised ML algorithms used in stock price prediction, including Linear Regression, Support Vector Regression (SVR), Decision Trees, Random Forest (RF), Gradient Boosting methods such as XGBoost, and Neural Network-based models. The study examines the capabilities of these algorithms in capturing nonlinear market behaviors, handling noise, and improving prediction accuracy.

Key factors such as feature engineering, technical indicators, preprocessing strategies, and hyperparameter optimization are discussed to highlight their role in enhancing model performance. A comparative assessment from existing literature indicates that ensemble learning methods and deep neural models outperform traditional statistical approaches in forecasting accuracy and robustness. Despite significant progress, challenges such as market volatility, data sparsity, and susceptibility to overfitting persist. This review concludes that while supervised ML techniques cannot completely eliminate market unpredictability, they substantially enhance forecasting reliability and serve as effective tools for financial analysis and data-driven decision-making.