Data Science & Analytics
Data Science & Analytics
Foodgrain Forecast
15 years of national grain data, forecast and audited.
2025
Highlights
- → 4-step SQL pipeline: import, deduplication, indexing, multi-level aggregation rollups
- → Prophet models for 30/90-day forward forecasting with ±8–12% uncertainty bands
- → Z-score and rolling-window anomaly detection — Punjab, Jharkhand, Chhattisgarh flagged most
- → 5-page Power BI dashboard, 30+ DAX measures, state-level choropleth
2.2M+
records
500+
anomalies flagged
15
years of data
Stack
PythonProphetSQLPower BIDAX