from typing import Callable
import pandas as pd
from .building_heating_profile import WEIGHT_NAME_REQUIRED
[docs]
def month_length_proportionnal_weight(dates: pd.DatetimeIndex) -> pd.Series:
"""Create a Series attributing a weight to each datetime of the index
the weight depends only of month and year of the datetime.
The weight attributed to each datetime equals month length over year length
Args:
dates (pd.DatetimeIndex): DatetimeIndex (a month can appear multiple times)
Returns:
pd.Series: Series with correct format to be used as weight
"""
return pd.Series(
dates.daysinmonth / (365 + dates.is_leap_year), index=dates, name=WEIGHT_NAME_REQUIRED
)
[docs]
def day_length_proportionnal_weight(dates: pd.DatetimeIndex) -> pd.Series:
"""Create a Series attributing a weight to each datetime of the index
the weight depends only of date and month of the datetime.
The weight attributed to each datetime equals day length over month length
Args:
dates (pd.DatetimeIndex): DatetimeIndex (a date can appear multiple times)
Returns:
pd.Series: Series with correct format to be used as weight
"""
return pd.Series(1 / (dates.daysinmonth), index=dates, name=WEIGHT_NAME_REQUIRED)
[docs]
def apply_hourly_pattern(
hourly_index: pd.DatetimeIndex, hourly_mapping: dict[int, float]
) -> pd.DataFrame:
"""Provide a DataFrame with correct weight format to disaggregate daily load into hourly load with a daily pattern.
Args:
hourly_index (pd.DatetimeIndex): Index
hourly_mapping (dict[int,float]): assiossate to each hour (int between 1 and 24) a weight (sum should equals one)
Returns:
pd.DataFrame: DataFrame weight correct format
"""
return pd.DataFrame(
hourly_index.hour.map(hourly_mapping),
index=hourly_index,
columns=[WEIGHT_NAME_REQUIRED],
)
[docs]
def apply_weekly_hourly_pattern(
hourly_index: pd.DatetimeIndex, hourly_mapping: Callable[[int, int], float]
) -> pd.Series:
"""Provide a Series with correct weight format to disaggregate daily load into hourly load with a hourlt weekly pattern.
It is better to have weight sum on 24 hours equals to 1 instead of weight sum equals to 1 over a week because week are note always complete in a month. This can lead to false disaggregation.
Args:
hourly_index (pd.DatetimeIndex): Index
hourly_mapping (dict[tuple[int,int],float]): assiossate to each hour and each day of week (int between 1 and 7, int between 1 and 24) a weight (sum should equals one over each day)
Returns:
pd.Series: Series weight correct format
"""
return (
pd.Series(hourly_index, index=hourly_index)
.apply(lambda x: hourly_mapping(x.dayofweek, x.hour))
.rename(WEIGHT_NAME_REQUIRED)
)