How jarvis3954p/lna_gainnoise-predictor works
Project type and purpose
• This is a full-stack machine-learning application (“LNA Performance Predictor”) that spans data ingestion, EDA, model training/evaluation, artifact serialization, and deployment via a Flask-based web interface.
• Purpose: Given LNA material, architecture, frequency, bandwidth, predict Gain and Noise Figure by loading a trained Gradient Boosting model.
File structure
• Top-level directories/files:
• dataset.csv (raw data)
• processed_lna_dataset.csv (cleaned/preprocessed)
• eda.ipynb (exploratory data analysis)
• model.ipynb (model training & comparison)
• prediction_app.py (wrapper to load models & scalers, perform inference)
• web_app.py (Flask application for UI & REST endpoints)
• *.pkl (model artifacts: gb_model_gain.pkl, gb_model_noise.pkl, scaler.pkl, encoder.pkl)
• requirements.txt (dependency list)
• Monolithic repo—no microservices. ML pipeline and web server co-exist.
README for additional insights
• ML pipeline stages: EDA → preprocessing → train/test split → train 3 regressors → evaluate (MAE, R²) → select champion model → serialize artifacts.
• Deployment: Flask app exposes a form, calls prediction_app to load .pkl artifacts, returns prediction and visualizations (Matplotlib charts).