Consumption and demand forecasts

Choose the measurement basis before reporting forecast accuracy: rt-cons is real-time consumption, uecm is settlement consumption, and load-plan is the demand forecast. Publication timing varies; inspect availability rather than promising a fixed delay.

from eptr2 import EPTR2
from eptr2.composite import get_hourly_consumption_and_forecast_data

eptr = EPTR2(use_dotenv=True, recycle_tgt=True, strict_params=True)
df = get_hourly_consumption_and_forecast_data(
    start_date="2026-07-01", end_date="2026-07-01", eptr=eptr,
)
print(df[["dt", "load_plan", "rt_cons", "consumption"]])

consumption uses UEÇM where present and otherwise real-time consumption; uecm may be absent when none is available. This is a fallback series, not uniformly settled demand. Preserve its source in downstream analysis. Hourly quantities are MWh and delivery timestamps use Europe/Istanbul.

Check completeness and duplicate timestamps before aggregating. Do not fill missing demand with zero. For forecast errors, state the actual-data basis and the denominator; avoid percentage division by zero.

Read examples.md for a source-aware accuracy calculation. Use whichever Python or CLI execution facility the user's runtime provides.