COVID-19 Sentiment Clustering & Classifier Model
- A Twitter-based clustering and classification model able to determine the sentiment polarity (Positive, Neutral, Negative) of Tweets.
- The clustering model implemented is based on the Modified K-Means algorithm.
- The classifier models are Supervised Machine Learning algorithms consisting of Naïve Bayes, Support Vector Machine, Logistic Regression, Random Forest, and Combination Ensemble (built from the aforementioned individual classifiers).
- Python 3
- To run main.py and webapp.py, the requirements are installed through:
python -m pip install -r requirements.txt- To train the model and run the system.ipynb, the requirements are installed through:
python -m pip install -r requirements_training.txt- cluster.py -> Python file that accepts sentiment inputs and returns the predicted cluster by the Modified K-Means algorithm.
- classify.py -> Python file that accepts sentiment inputs and returns the predicted sentiment polarity (Positive, Neutral, Negative) as classified by the classifier models.
- webapp.py -> Web application utilizing the Flask framework. Likewise accepts the sentiment inputs and returns the predicted cluster and sentiment polarity.
- system.ipynb -> Jupyter Notebook file documenting the project. Includes the clustering and classification of inputs, as well as training the system.