Andrianarimanana/Storm_Prediction_Madagascar

This repository provides tools and resources for predicting thunderstorms in northern Madagascar based on Meteosat Second Generation Data on Cloud-Top Temperature in order to climate risk management. # Storm prediction in north of Madagascar Welcome to the **Storm prediction in north of Madagascar** project! This repository provides tools and resources for predicting thunderstorms in northern Madagascar based on Meteosat Second Generation Data on Cloud-Top Temperature in order to climate risk management. ## Table of Contents 1. Overview 2. Dataset 3. Setup Instructions ## Overview The porpuse of this project is a machine learning focused on forcasting thunderstorms in northern Madagascar, particularly around Nosy Be. The project aims to provide accurate short-term predictions (0–6 hours) to mitigate risks, protect lives, and support emergency responses in this vulnerable region. ## Dataset The repository includes data derived from EUMETSAT's Meteosat Second Generation satellite observations , focusing on cloud-top temperature. Storms are detected in real-time using the 2D-wavelet transform method by Klein et al. (2018), with database creation contributed by Rakotomanga et al. (2024). Dataset Details: Inputs: Each row includes observation time (year, month, day, hour, minute), geographic coordinates (latitude and longitude), and storm characteristics (intensity, size, and distance). Labels: Training data includes binary labels for storm occurrence predictions at 1-hour (Storm_NosyBe_1h) and 3-hour (Storm_NosyBe_3h) lead times. Time Span: Data spans November to April from 2004–2019 for training and 2020–2024 for testing. Files in Repository: train.csv: Contains input features and labels for model training. test.csv: Contains input features without labels for model evaluation. ## Setup Instructions To set up this project locally, follow these steps: 1. Clone the repository: ```bash git clone github.com 2. Activate virtual environment (make sure pipenv is already installed): ```bash pipenv shell 3. Install Dependencies: ```bash pipenv install 4. Activate the Virtual Environment ```bash pipenv s …