How shivamshinde123/thyroid_disease_detection_internship works
Project type and purpose
• This is an end-to-end machine learning application for thyroid disease detection.
• It includes a data pipeline (data ingestion, validation, preprocessing), model training with MLflow tracking, and a Streamlit-based web application for serving predictions.
File structure
• Top-level data directories:
• Data/Raw_data – stores raw CSV from UCI
• Data/Processed_Data – stores cleaned train/validation splits
• Preprocess_pipeline – stores serialized transformers (label encoder, preprocessing pipeline)
• Models – stores the trained model artifact (model.joblib)
• Logs – per-stage log files
• Metrics – evaluation outputs (classification_report.csv, metrics.json)
• mlruns – MLflow experiment artifacts and metrics
• src – contains all Python code:
• step0_utility_functions, step1_make_dataset, step2_data_validation, step3/4 preprocessing stages, step5_model_training
• webapp.py (Streamlit front end)
• tests for each pipeline stage
• config files: params.yaml, requirements.txt, .streamlit/config.toml
README for additional insights
• Describes project flow: data ingestion → validation → preprocessing → training/evaluation → web deployment
• Lists technologies: Python, Pandas, NumPy, scikit-learn, MLflow, Pydantic, Streamlit