WeatherIQ — ML Inference System

WeatherIQ is a machine learning inference system for classifying weather conditions from environmental parameters.

The project focuses on deploying machine learning models through a modular inference architecture with:

• API-based model serving
• reproducible preprocessing pipelines
• frontend/backend integration
• containerized deployment
• real-time prediction workflows

The platform exposes a FastAPI inference service alongside an interactive Streamlit frontend for live predictions.

---

Live Demo

Frontend Application

🌐 https://cloud-cover-classification-cm3ajlcsp6ytwzkdphchzy.streamlit.app/

Backend API & Documentation

API Base URL

https://cloud-cover-api.onrender.com

Swagger Documentation

https://cloud-cover-api.onrender.com/docs

---

System Architecture

┌──────────────────────┐
│   Streamlit Client   │
└──────────┬───────────┘
           │
           ▼
     POST /predict
           │
           ▼
┌──────────────────────┐
│    FastAPI Backend   │
└──────────┬───────────┘
           │
           ▼
┌──────────────────────┐
│  ML Inference Layer  │
│                      │
│ • Feature Scaling    │
│ • One-Hot Encoding   │
│ • Classification     │
└──────────┬───────────┘
           │
           ▼
 Prediction + Confidence

---

Engineering Highlights

• RESTful inference API using FastAPI
• Request validation using Pydantic schemas
• Unified preprocessing and inference workflow
• Serialized ML artifact bundle for reproducible inference
• Probability-based confidence estimation
• Decoupled frontend and backend services
• Containerized deployment using Docker and Docker Compose
• Modular deployment-ready project structure

---

API Contract

Endpoint

POST /predict

---

Request Example

{
  "Temperature": 14.0,
  "Humidity": 73,
  "Wind_Speed": 9.5,
  "Precipitation": 82.0,
  "Atmospheric_Pressure": 1010.82,
  "Visibility": 3.5,
  "Season": "Winter",
  "Location": "inland"
}

---

Response Example

{
  "prediction": "Rainy",
  "confidence": 0.87
}

---

ML Inference Pipeline

The backend loads a serialized artifact bundle containing:

• trained classification model
• feature scaler
• one-hot encoder
• feature mappings

This ensures preprocessing consistency between training and inference environments while reducing feature mismatch during deployment.

---

Technology Stack

Layer	Technologies
Machine Learning	scikit-learn
Backend	FastAPI, Uvicorn
Frontend	Streamlit
Data Processing	pandas, NumPy
Containerization	Docker, Docker Compose
Deployment	Render, Streamlit Cloud

---

Dockerized Deployment

Build Containers

docker compose build

Run Services

docker compose up

---

Local Development

Clone Repository

git clone https://github.com/YuvarajGML/Cloud-Cover-Classification.git
cd Cloud-Cover-Classification

---

Start Backend

cd backend
pip install -r requirements.txt
uvicorn app:app --reload --port 10000

---

Start Frontend

cd Frontend
pip install -r requirements.txt
streamlit run frontend.py

---

Repository Structure

Cloud-Cover-Classification/
│
├── backend/
│   ├── app.py
│   ├── requirements.txt
│   ├── Dockerfile
│   └── weather_model_bundle.pkl
│
├── Frontend/
│   ├── frontend.py
│   ├── requirements.txt
│   └── Dockerfile
│
├── docker-compose.yml
├── .dockerignore
└── README.md

---

Author

Yuvaraj Gopi