This project involves the classification of tweets about the Coronavirus (COVID-19) as positive or negative. The dataset contains tweets categorized into five classes: 'Extremely Negative', 'Extremely Positive', 'Negative', 'Neutral', and 'Positive'. The objective is to filter out neutral tweets and convert the remaining tweets into binary classes: positive and negative.
The project is contained within a single Jupyter notebook:
TP_Exam.ipynb- This notebook contains the full pipeline of the project, including exploratory data analysis (EDA), data preprocessing, model training, evaluation, and interpretation.
The notebook follows the following steps:
- Exploratory Data Analysis (EDA)
- Data Cleaning
- Text Vectorization
- Models Training and Evaluation
- Hyperparameter Optimization
- Model Interpretability by using of LIME and SHAP
- Comments and Analysis:
- pandas
- seaborn
- numpy
- matplotlib
- scikit-learn
- lime
- shap
- spacy
- xgboost
- joblib
To run the project, ensure you have Jupyter Notebook installed and simply open the TP_Exam.ipynb file. The notebook can be executed cell-by-cell to reproduce the analysis and results.
This project is part of an academic examination. Contributions are not requested, but feedback and suggestions for improvement are always welcome.
This project is licensed under the terms of the MIT license.