Below are some of the data analysis, machine learning, and Power BI projects I've worked on.
Cryptocurrency Prediction
Developed a predictive model to forecast cryptocurrency price trends using historical data, advanced machine learning algorithms, and time-series analysis.
- Tech Stack: Python, Pandas, NumPy, scikit-learn, RaandomForestRegressor, Flask
- Focus: Price forecasting, feature engineering, model evaluation
- Outcome: Improved forecasting accuracy by applying Random forest Regressor model and built the application using flask.
Book Genre Classification
Built a text classification model to automatically label books by genre based on their descriptions and metadata, using NLP techniques.
- Tech Stack: Python, scikit-learn, NLP libraries (NLTK/spacy), Pandas, Flask
- Focus: Text preprocessing, vectorization (CountVectorizer, TF-IDF), classification algorithms
- Outcome: Achieved high accuracy in predicting genres, demonstrating the model’s potential for automated categorisation.
Heart Attack Prediction
Leveraged machine learning algorithms to predict the likelihood of heart attacks based on clinical and laboratory data, improving early detection and healthcare outcomes.
- Tech Stack: Python (Pandas, NumPy, scikit-learn), Jupyter Notebook, Streamlit
- Focus: Classification (Random Forest, Logistic Regression), model evaluation (ROC, AUC)
- Outcome: Provided actionable insights for healthcare professionals by identifying high-risk patients with significant accuracy.
Netflix Recommendation System
Created a content-based filtering system to suggest movies and TV shows based on user preferences and metadata similarity, enhancing the user experience.
- Tech Stack: Python, Pandas, scikit-learn, CountVectorizer
- Focus: Cosine similarity, content-based filtering, feature extraction
- Outcome: Delivered relevant recommendations by analysing plot overviews, genres, and user ratings.
Total Call Centre Cases Forecast
Implemented SARIMA and Prophet models to predict call volumes for a call centre, enabling resource optimisation and improved staffing.
- Tech Stack: Python, Pandas, NumPy, statsmodels, Prophet, SARIMA
- Focus: Time-series forecasting, model selection (SARIMA vs. Prophet), hyperparameter tuning
- Outcome: Increased forecasting accuracy and allowed management to reduce wait times by better scheduling staff.
Credit Card Fraud Detection
Developed a classification model to detect fraudulent credit card transactions. Utilised advanced machine learning techniques and model comparison to accurately identify suspicious activity and reduce financial losses.
- Tech Stack: Python, scikit-learn, PyCaret, Flask
- Focus: Classification metrics, model comparison (e.g., Ensemble methods, Random Forest, LightGBM), feature engineering
- Outcome: Achieved high accuracy in identifying fraudulent transactions, significantly reducing false positives and enhancing financial security.
