AI-Powered Predictive Analytics for Fuel Price Forecasting

FuelCast is a state-of-the-art dashboard that leverages hybrid machine learning models (SARIMAX + XGBoost) to forecast US gasoline prices. It provides real-time insights, model performance metrics, and explainable AI features to help stakeholders make data-driven decisions.

🚀 Advanced Forecasting

• Hybrid AI Model (SARIMAX + XGBoost)
• Real-time price predictions
• Interactive time-range filtering
• Responsive Recharts components

🎨 Premium Design

• Dark mode optimized
• Glassmorphism effects
• Smooth micro-animations
• Vibrant gradient accents

• Framework: Next.js 14+ (App Router)
• Language: TypeScript (Strict mode)
• Charts: Recharts
• Styling: Tailwind CSS
• Icons: Lucide React
• Backend: FastAPI (Python)

• Node.js 18+ installed
• FastAPI backend running on http://localhost:8000

# Install dependencies
npm install

# Run development server
npm run dev

# Build for production
npm run build
npm start

Open http://localhost:3000 in your browser.

The dashboard expects the following FastAPI endpoints:

Returns forecast data points:

[
  {"date": "2024-01-01", "actual": 3.45, "sarima": 3.44, "xgboost": 3.46},
  {"date": "2024-01-08", "actual": null, "sarima": 3.50, "xgboost": 3.52}
]

Returns model performance metrics:

{
  "sarima_rmse": 0.04,
  "xgboost_rmse": 0.02,
  "current_price": 3.45
}

Returns feature importance scores:

[
  {"feature": "Crude Price Lag-4", "score": 0.8},
  {"feature": "Seasonality", "score": 0.6}
]

FuelCast/
├── app/
│   ├── layout.tsx          # Root layout
│   ├── page.tsx            # Home page
│   └── globals.css         # Global styles
├── components/
│   ├── Dashboard.tsx       # Main dashboard component
│   ├── KPICards.tsx        # Metrics cards
│   ├── ForecastChart.tsx   # Line chart
│   ├── FeatureImportance.tsx # Bar chart
│   └── TimeRangeSelector.tsx # Filter control
├── lib/
│   ├── api.ts              # API client
│   └── utils.ts            # Utility functions
└── package.json

• All Time: Shows complete historical and forecast data
• Last Year: Displays data from the past 12 months
• Forecast Only: Shows only future predictions

MIT