CaReCur/src/features/label_events.py

"""The purpose of this script is to add a binning to define "buckets" of curtailment events at different cutoffs.

Each of these will be used in a logistic regression to test different sensitivities.

To start, curtailment events will be defined as cutoff values in daily curtailment events.
"""
import pandas as pd

from loguru import logger
from src.conf import settings

OUTPUT_DIR = settings.DATA_DIR / "processed/training/"


if __name__ == "__main__":

    data = pd.read_parquet(OUTPUT_DIR / "0_labeled_data_daily.parquet")

    cutoffs = [
        0.01,
        0.03,
        0.05,
        0.1,
    ]

    for cutoff in cutoffs:
        logger.debug("Labeling curtailment events with {cutoff} pct curtailed of total production.", cutoff=cutoff)
        data[f"curtailment_event_{cutoff:.2f}"] = (
            data["solar_curtailment"] / data["solar"] > cutoff
        )

    data.to_parquet(OUTPUT_DIR / "1_labeled_curtailment_events.parquet", index=False)