Sentiment Analysis On Masculinity Saturday

Project Description

The Amerix movement has gained popularity in Kenya [and beyond] over the recent years. It was initiated in the year 2019 by [Eric Amunga], and has grown to thousands of followers across select social media platforms. The movements key points of focus are Mens Health and wellness & reproductive health. The Agenda & Philosophy of the movement is mainly fronted through key hashtags and taglines, mainly: #MasculinitySaturday, #ManDay, #FoodFriday, #BetterTogether, Change or Perish, and so on.

This project's key aim to carry out sentiment analysis on the tweets associated with this movement (using the above mentioned taglines), and make reasonable, analytical inferences on the impact of the movement since its inception. The project deploys top-level Machine Learning and Deep Learning models and architectures to make the best possible predictions on whether tweets (and their implications) are Positive, Negative, or Neutral, the volume of positivity, negativity, or neutrality and the possible implied socialistic meanings [and connotations].

Key Models & Architectures

A. Machine Learning

1. Gaussian Naive Bayesian (GNB)
2. Random Forests (RF)
3. Extreme Gradient Boosters (XGB)
4. Categorical Boosters (CatBoost)

B. Deep Learning

5. Bert Transfomers
6. Roberta Transformers
7. Twitter-Roberta
8. TwitterNLP

F1 Scores

1. Bert == 0.97
2. Roberta == 0.96
3. RF == 0.90
4. CatBoost == 0.83
5. XGB == 0.80
6. GNB == 0.78
7. Twitter-Roberta == (pending)
8. TWitterNLP == (pending)