Research on user sentiment in social media is essential for understanding user opinions. Classification accuracy is raised by a novel sentiment analysis approach combining sentiment mapping of emoticons and hashtags with text-based machine learning models. Our methodology substitutes TF-IDF vectorization and a Random Forest classifier for deep learning, which calls for massive datasets and processing resources, for effective text-based sentiment analysis. We employ hashtag-based sentiment categorization and predefined emoji sentiment ratings to improve contextual awareness. With an accuracy of 89.87%, macro-averaged precision, recall, and F1-score of 90%, the proposed model performs nicely over sentiment categories. Attitudes are correctly classified by the Confusion Matrix, ROC-AUC Curve, and Swarm Plot. The model has good discrimination shown by its 0.98 ROC-AUC score. Although closely related emotions like Neutral vs. Positive and Fearful vs. Sad have little misclassifications, the stability and dependability of the model in practical applications are verified. These findings augment previous research on hybrid sentiment analysis systems utilizing sentiment detection through emoji- and hashtag-based methodologies. Random Forest classifiers perform conventional machine learning models in sentiment analysis, as shown by the results. The proposed method is useful for social media monitoring, customer sentiment analysis, and opinion mining due to its precision, resource effectiveness, and scalability. The model achieves an accuracy of 89.87% and a ROC-AUC score of 0.98. A comparison using SVM and Naïve Bayes reveals enhanced performance in both precision and recall, showing the model’s robustness and resource efficiency for real-time sentiment analysis in social media platforms.