README.md

metadata

title: rainfall
emoji: 👀
colorFrom: purple
colorTo: red
sdk: gradio
sdk_version: 5.27.1
app_file: app.py
pinned: false
hf_oauth: true

Rainfall Prediction with LSTM and Gemma 2 Interpretation

This application predicts daily rainfall based on weather features (temperature, humidity, and sunshine duration) using an LSTM model, and provides a natural language interpretation of the prediction using the Gemma 2-27B language model from Google via Hugging Face Inference API.

---

📊 Features

• ✅ Predicts daily rainfall (mm) from 3 key weather inputs.
• 🧠 Uses an LSTM neural network trained on historical weather data.
• 💬 Generates automatic explanation and recommendations using Gemma 2-27B (google/gemma-2-27b-it) via Hugging Face InferenceClient.
• 🌐 Built with Gradio UI for easy interaction.

---

🧪 Input Parameters

Parameter	Description
Average Temperature (°C)	Daily mean temperature in Celsius
Relative Humidity (%)	Average daily relative humidity
Sunshine Duration (hours)	Total sunlight duration during the day

---

🧮 Output

• Predicted Rainfall (mm): Estimated rainfall amount based on inputs.
• Interpretation: AI-generated weather explanation using LLM (Gemma 2-27B), including suggestions for outdoor activities.

---

🚀 Getting Started (for Deployment)

1. Prepare the required files:

   • app.py — main app
   • retrained_lstm_model.h5 — trained LSTM model
   • scaler_X.joblib, scaler_y.joblib — scalers used for input/output normalization

2. Add Hugging Face token to Space secrets:

   • Go to your Hugging Face Space → Settings > Secrets
   • Add a secret:
     • Name: HF_TOKEN
     • Value: your Hugging Face API token

3. Install dependencies using requirements.txt:

   tensorflow>=2.8.0
   numpy>=1.19.5
   pandas>=1.3.5
   scikit-learn>=1.0.2
   joblib>=1.1.0
   gradio>=3.20.0
   huggingface_hub>=0.18.0
   

Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference