Predicting stock prices is a key challenge in finance, influencing decisions made by investors and traders. Traditional forecasting methods often struggle to capture the complex and unpredictable nature of stock market movements. This study explores how machine learning models—Random Forest (RF), Support Vector Machines (SVM), Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU)—perform in stock price prediction. The models were evaluated based on accuracy, interpretability, and computational efficiency. Using Tata Motors' stock price data from 2014 to 2024 (sourced from Yahoo Finance), we applied feature engineering techniques such as moving averages, Relative Strength Index (RSI), and Bollinger Bands to improve predictions. Data preprocessing, including handling missing values and normalization, was carried out to enhance model performance. Our findings show that Random Forest delivered the best results, with the lowest RMSE (0.23), making it the most effective model for stock price forecasting. SVM performed moderately well but fell short in accuracy. LSTM and GRU had higher error rates, suggesting they require extensive fine-tuning and larger datasets for optimal performance. Overall, Random Forest proved to be the most reliable and interpretable model for stock prediction. Future research can explore hybrid approaches and integrate sentiment analysis from financial news and social media to further improve prediction accuracy.