This research paper presents a comprehensive analysis of student performance based on a merged dataset from two subjects: Mathematics and Portuguese. The dataset encompasses a wide range of student attributes and behaviours, allowing us to explore various factors influencing academic success. The study employs exploratory data analysis, data visualization, statistical tests, and machine learning techniques to uncover patterns and relationships within the data. The analysis begins with data preprocessing, including column renaming, categorical variable creation, and merging of datasets. Visualizations such as count plots, heatmaps, and distribution plots are utilized to understand the distribution of final grades and correlations between attributes. Furthermore, the study investigates the impact of romantic status, weekend alcohol consumption, parental education, social activities, desire for higher education, and urban/rural living areas on student performance. Chi-square tests for association are conducted to assess the significance of relationships between categorical variables and final grades. Machine learning models, including Decision Trees, Random Forests, Support Vector Machines, Logistic Regression, AdaBoost, and Stochastic Gradient Descent, are employed to predict student performance. Feature selection techniques are also employed to determine the optimal number of features for achieving high prediction accuracy. Through this comprehensive analysis, we gain insights into the complex interplay of various factors affecting student performance. The findings contribute to a deeper understanding of academic success and highlight potential areas for intervention and improvement. This research serves as a valuable resource for educators, policymakers, and researchers seeking to enhance educational outcomes and facilitate targeted interventions for students.