- Analyzed the Pfizer BioNTech vaccine tweets to assess public opinions and acceptance.
- Recommended that showing the scientific data, information about the approvals and efficacy of the vaccine, and addressing concerns like safety, severe side effects, and allergies in ads and campaigns would reduce public hesitancy toward the vaccine.
- Analyzed the banking data and developed ML models to detect transaction patterns that signal high risk for fraud, especially for older and vulnerable adult customers, which scammers could target.
- Achieved a maximum AUC score of 0.9694 using a Decision Tree Classifier which turned out to be the best model.
- Analyzed spam data and developed three pipelines using TF-IDF vectors, manual features, and combined TF-IDF vectors. The evaluation metric used was F0.5 giving slightly more importance to precision than recall for this imbalanced dataset.
- Achieved a maximum F0.5 score of 0.9458 using an XGBoost Classifier which turned out to be the best model.
Attached custom preprocessor is used.
Manual features are extracted and used these as the input to our XGBoost Classification model.
- The 2020 Chicago crime data was visualised through interactive dashboard in R-Shiny App.