Data Sources

Every dataset, API and formula the platform relies on — with live status and source classification. No source is silently substituted.

Data source classifications

OFFICIAL_SOURCEHighest authority

Physical models and regulatory
formulas with published references.
→ pvlib (GHI closure, POA, PVWatts)
→ CERC/KERC tariff rules

DATASET_DERIVEDReal-world data

Real datasets downloaded from
Kaggle, OpenStreetMap, NLR NSRDB.
→ Anikannal solar plant data
→ OSM substations (344 in Bengaluru)
→ NSRDB Himawari/SUNY India

MODEL_LEARNEDTrained on real data

ML model predictions trained on
real data.
→ solar_forecast_model.pkl (GHI)
→ cloud_risk_classifier.pkl (P(cloud))
→ Kaggle PV model (HistGBT)

FALLBACK_DEFAULTDocumented, overridable

Transparent heuristics used when
no model or data is available.
→ confidence = 1 − 0.35·cloud
→ γ = −0.0035/°C (c-Si default)
→ Dynamic risk weights (0.6/0.4)

Provenance: REAL_BENGALURU + REAL_INDIA + REAL_COORDINATE_BASED — The platform distinguishes data sourced from real Bengaluru coordinates (weather, substations) from India-wide datasets (Kaggle PV) and coordinate-based lookups (Open-Meteo at any lat/lon).

Checking backend…