This research work assesses the effectiveness of a Random Forest and Naive Bayes ensemble in addressing the challenging task of email categorization. In order to guarantee the validity of the analysis utilizing actual email data, the research applies crucial preprocessing techniques including feature selection and data integrity checks in addition to machine learning models. The ensemble model, which is a combination of Random Forest and Naive Bayes, is trained and evaluated with an emphasis on important performance metrics including accuracy and classification reports. In order to handle frequent issues with email data, such missing values, robust approaches are used. Specifically, the Voting Classifier shows itself to be a potent instrument that improves overall model performance by offering a fair way to classify emails. The findings provide an extensive analysis of memory, accuracy, and precision together with a comprehensible depiction using confusion matrices. This work emphasizes the importance of ensemble learning and its potential in addressing algorithmic trade-offs, beyond its technical contributions. The research contributes significant insights to discussions on effective and dynamic email categorization by illuminating the subtle dynamics of email filtering techniques. The work functions as a foundational component by offering practitioners and academics instructional value in addition to giving immediate data. It establishes the foundation for further developments in this important area and promotes a better comprehension of the advantages of integrating various machine learning approaches for changing email categorization problems. In this research, we evaluate the performance of various classification algorithms, including a Voting Classifier, K-Nearest Neighbors, Gaussian Naive Bayes, and Random Forest, on a given dataset. The Voting Classifier demonstrates high accuracy (95.9%) and overall superior performance with notable precision (99%), recall (89%), and F1-Score (95%). K-Nearest Neighbors achieves moderate accuracy (80.2%) but exhibits lower precision (63%) and F1-Score (69%). Gaussian Naive Bayes and Random Forest both yield commendable accuracies (93.6% and 93.7%, respectively) with competitive precision, recall, and F1-Score metrics. This study provides valuable insights into the comparative strengths and weaknesses of these algorithms, offering a comprehensive perspective for practical applications in classification tasks.

Index Terms- Ensemble learning, Random Forest, Naive Bayes, Voting Classifier, email categorization, classification tasks.