Skip to content

Two-stage XGBoost model for predicting cloud-to-ground lightning: classify occurrence and estimate strike counts, deployed via a Streamlit web app.

Notifications You must be signed in to change notification settings

ArjunaRey/LightningPrediction

Repository files navigation

This project implements a two-stage machine learning model to predict cloud-to-ground (CG) lightning using ERA5 atmospheric data and actual lightning strike records. The first stage classifies whether lightning will occur, and the second stage estimates the number of strikes if detected. The model is evaluated using F1-Score, Accuartion, Precision, Recall, CSI, MAE, RMSE, and NRMSE, and deployed as a Streamlit web app for real-time predictions.

About

Two-stage XGBoost model for predicting cloud-to-ground lightning: classify occurrence and estimate strike counts, deployed via a Streamlit web app.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages