Market trend predictions play a crucial role for investors and policymakers along with financial analysts because they produce well-informed choices in an unpredictable market. Stock market prediction methods using conventional strategies mainly analyze opening and closing prices while working with restricted datasets and utilizing restricted machine learning models that limit their universal application and successful results. A study has been designed to overcome existing gaps through the integration of exploratory data analysis (EDA) and Time Series Analysis (TSA) together with Autoregressive Integrated Moving Average (ARIMA) along with XGBoost, Random Forest, Decision Tree and Linear Regression machine learning models. The main concentration of this investigation points toward regression-based models which show a shortage in earlier research work. Multiple predictive frameworks emerge from this work since it uses historical data and multiple model applications for achieving enhanced accuracy. The incorporation of EDA and TSA provides better insights into data patterns that leads to enhanced prediction model interpretability together with improved performance levels. The study shows how the chosen models perform in relation to each other while showing that regression-based techniques matter the most for forecasting methods which aim to support contemporary financial market requirements.