📖 Documentation

1. Getting Started

The Cardamom Procurement Intelligence System is a static web application that runs entirely in your browser. No server or backend is required — all forecasting data is pre-computed.

Quick Start

1. Open index.html in a modern browser (Chrome, Firefox, Edge, Safari)
2. Click "Load Sample Data" or upload your own CSV
3. Click "Generate 30-Day AI Forecast"
4. Explore analytics, calculate costs, and export reports

2. Data Loading

Sample Data

Click the "Load Sample Data" button to load the built-in dataset of 2,869 rows spanning 2017–2025. This provides instant access to forecasting without any file uploads.

Custom CSV Upload

Upload your own CSV file with all 39 features (or at minimum time and Avg.Price (Rs./Kg)). Missing features are auto-filled with intelligent defaults.

time, Avg.Price (Rs./Kg), MaxPrice (Rs./Kg), Total Qty Arrived (Kgs), ...

Minimum 30 rows are required. The system accepts any date range and generates live 30-day predictions starting from the day after your last data point.

3. Generating Forecasts

The "Generate 30-Day AI Forecast" button loads pre-computed predictions from three models:

• TimesFM: Trend and seasonal components
• LSTM-CNN: Risk classification and volatility estimation
• XGBoost: Binary buy/sell signals

Results include daily price predictions, percentage changes, risk levels, and actionable recommendations.

4. Analytics Dashboard

Four visualization tabs provide different perspectives on the forecast:

• Price Trend: Line chart with daily % bars and optimal entry marker
• Risk Calendar: Day-by-day risk visualization
• Recommendation Breakdown: Donut chart of recommendation distribution
• Confidence Gauge: Average model confidence meter

5. Cost Calculator

Enter your desired quantity (kg) using the input field or slider (100–10,000 kg). The calculator compares:

• Cost Today: Current market price × quantity
• Optimal Cost: Best entry price × quantity
• Savings: Difference in absolute and percentage terms

6. What-If Analysis

Adjust 9 parameters to simulate different market scenarios:

• Temperature (°C), Precipitation Lag (mm), Soil Moisture (0-1)
• Quantity Arrived (kg), Quantity Sold (kg), Max Price (₹/kg)
• Market Open (yes/no), Flood Event (yes/no), Lockdown (yes/no)

Preset Scenarios

Four built-in presets: Monsoon, Supply Shortage, Optimal Conditions, Crisis Mode.

7. Exporting Data

• CSV: Full 30-day forecast table with all columns
• PDF: 4-page executive report with summary, charts, and recommendations
• Share Link: Copies URL with current settings (currency, quantity)

8. Currency Support

Three currencies available via the top-right selector:

• USD ($) — Base currency
• SAR (ر.س) — Saudi Riyal at 3.75
• INR (₹) — Indian Rupee at 83.50

All prices, charts, and exports update live when switching currencies.

9. Deployment

GitHub Pages

1. Push the website/ folder to your repository
2. Go to Settings → Pages
3. Set Source to "Deploy from a branch"
4. Select the branch and /website folder
5. Save — site goes live at username.github.io/repo/website/

Local Development

# Python
python -m http.server 8000 --directory website

# Node.js
npx serve website

10. Data API

To update forecasts, replace assets/data/forecasts.json with fresh predictions from the Python backend. Run:

python generate_forecasts.py --output website/assets/data/forecasts.json

The JSON schema must match the structure documented in forecasts.json.