Publication Date

Spring 2024

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)


Computer Science

First Advisor

Robert Chun

Second Advisor

Fabio Di Troia

Third Advisor

Lokesh Ponnada


Instagram data, engagement class, sentiment analysis, RF, Stacking Classifier, XGB, SMOTE


With enormous amount of social media content, we can draw valuable insights. In this paper, we apply different Machine Learning and Deep Learning techniques on Instagram data to determine the techniques that work well to discover the engagement class of a social media post. Out of all the social media platforms, Instagram is growing rapidly not just in the number of users but also in terms of Advertisement and marketing surpassing YouTube’s advertisement revenue. The end goal of this paper is to propose a technique to predict the engagement class. We applied Random Forest (RF), Stacking Classifier, Extreme Gradient Boost (XGB), Voting Classifier, K- Nearest Neighbor (KNN), Neural networks for determining the engagement of a particular post. We check the effect of word embeddings like BERT, Word2Vec in predicting the engagement class. We also check the effect of applying SMOTE on the dataset. Secondly, the paper deals with finding caption sentiments using GPT model and lexicon-based (Vander) sentiment analyzer.

Available for download on Thursday, May 22, 2025