This project involves developing a machine learning-based classification system to predict cancer risk categories using relevant features from a structured dataset. The implementation follows an end-to-end workflow encompassing problem definition, data understanding, exploratory data analysis, data preprocessing, model development, optimization, evaluation, and deployment.
The dataset is analyzed to understand feature distributions, identify potential data-quality issues, and investigate relationships between input variables and the target classification. Multiple model-building experiments are conducted, including the use of XGBoost. Optuna is incorporated into the experimentation workflow to explore hyperparameter configurations and identify settings that may improve model performance.
The selected model is evaluated using appropriate classification metrics and saved for integration into a Streamlit-based web application. The application provides a user interface for entering the required features and obtaining predictions from the trained model. The project also includes model saving and deployment stages to demonstrate how a machine learning solution can be converted into an accessible application.
Overall, the project demonstrates the practical application of machine learning, exploratory data analysis, hyperparameter optimization, predictive modeling, and web application development in a healthcare-related classification task.