This paper investigates the efficacy of various classifiers for Emotion detection on unstructured data tweet text data. We compare the performance of traditional algorithms, including Decision Trees, K-Nearest Neighbors (KNN), and Naïve Bayes, along with their bagged versions, to assess their accuracy and robustness in classifying sentiments. The classifiers were evaluated using a dataset related to Emotion detection in tweets, with metrics including accuracy and Area Under the Curve (AUC). Our findings indicate that the Bagged Decision Tree classifier outperformed others, achieving an accuracy of 89.60% and an AUC of 0.81. The Naïve Bayes classifier also demonstrated strong performance, with an accuracy of 89.05% and an AUC of 0.81. Conversely, the Decision Tree classifier exhibited the lowest performance, highlighting the effectiveness of ensemble methods in enhancing classification accuracy. The results suggest that leveraging bagging techniques can significantly improve model performance, making them suitable for sentiment analysis tasks. This study contributes to the understanding of classifier performance in the context of unstructured data, offering insights for future research on advanced sentiment classification methodologies.