<b>A Survey Towards Effective Emotion Analysis: A Comparative Investigation of Machine Learning Models on Digital platform</b>

Emotion analysis in social-media dialogue is a discipline procedure that has ended up being obligatory in the contemporary academia. To successfully classify emotion, this paper conducts comparisons on different paradigms of canonical machine-learning including, Naive Bayes, LR, Support- Vector machine (SVM), Rand dom Forest, and K - nearest neighbours (KNN). Text in social -media was first processed by the more traditional preprocessing regimes and then transformed by a TF -IDF representation. Such adjudicated model outcomes by precision, F -score, correctness level and recall. The experimental results indicate that, SVM attained highest precision (90.3%), followed by LR (88.6%). Naive Bayes and random forests performed in a similar way with average results of KNN presenting a fairly low score. These findings allow concluding on the relevance of linear classifiers particularly SVM to high dimensional textual emotion problems and the article gives some advice on how to select the appropriate algorithms that should be used in emotion-analysis applications.