I built a crime rate prediction application in Python and Streamlit, taking it from SQL-based data querying through model deployment. It achieved 87% prediction accuracy on real-world crime data.
To improve the model, I addressed data inconsistencies through feature engineering and normalization, then tested and compared Random Forest and Linear Regression approaches with hyperparameter tuning. I also created visualizations of crime hotspots, temporal trends, and risk patterns.
My AI Interview Coach project uses FastAPI, MongoDB, and language model APIs to analyze resumes and job descriptions and generate interview questions. I added an answer evaluation system and adaptive follow-up questions, along with text- and speech-based interview modules.
Through Oracle Data Science Professional training, I completed hands-on work in data preprocessing, exploratory data analysis, machine learning, and model evaluation. I used Python, Pandas, NumPy, Matplotlib, and scikit-learn to construct and evaluate supervised models.
