祝日フラグの作成と連休の反映

import pandas as pd
import jpholiday

# 関数定義：祝日マートの作成
def get_holidays(start, end):
  '''
  start, str, YYmmdd
  end, sr, YYmmdd
  '''
  df = pd.DataFrame(pd.date_range(start, end), columns=['date'])
  # 年を除き、月日のみを抜粋
  df['mmdd'] = df['date'].dt.strftime('%m-%d')
  # 祝日フラグ
  df['holiday_flg'] = df['date'].map(jpholiday.is_holiday).astype(int)
  # 祝日名称
  df['holiday_name'] = df['date'].map(jpholiday.is_holiday_name)
  # 曜日
  df['dayname'] = df['date'].dt.day_name()

  return df

# 確認用に下記期間で祝日マート作成
# 2024年：8月の山の日とお盆が一体化
# 2026年：9月に国民の祝日がある
start = '20240101'
end = '20261231'
# 祝日マート作成
df_flg = get_holidays(start=start, end=end)
display(df_flg.head(2))
display(df_flg.tail(2))
# 確認用に祝日一覧取得
display(df_flg[df_flg['holiday_flg']==1].head(3))

# date	mmdd	holiday_flg	holiday_name	dayname
# 0	2024-01-01	01-01	1	元日	Monday
# 1	2024-01-02	01-02	0	None	Tuesday

# date	mmdd	holiday_flg	holiday_name	dayname
# 1094	2026-12-30	12-30	0	None	Wednesday
# 1095	2026-12-31	12-31	0	None	Thursday

# date	mmdd	holiday_flg	holiday_name	dayname
# 0	2024-01-01	01-01	1	元日	Monday
# 7	2024-01-08	01-08	1	成人の日	Monday
# 41	2024-02-11	02-11	1	建国記念の日	Sunday

# ###########################################
# 関数定義：前後２日分の祝日フラグ・名称を追加/更新
# ###########################################
def add_before_after_holidays(df):
  '''
  df: DataFrame, 祝日フラグ・名称を追加するdf
  '''
  # holiday_nameが欠損ではない日のholiday_flgは全て１にする
  df.loc[~df['holiday_name'].isna(), 'holiday_flg'] = 1

  # 前後２日分の祝日フラグ・名称を追加/更新
  df['holiday_flg_before'] = df['holiday_flg'].shift(1)
  df['holiday_name_before'] = df['holiday_name'].shift(1)
  df['holiday_flg_before2'] = df['holiday_flg'].shift(2)
  df['holiday_name_before2'] = df['holiday_name'].shift(2)

  df['holiday_flg_after'] = df['holiday_flg'].shift(-1)
  df['holiday_name_after'] = df['holiday_name'].shift(-1)
  df['holiday_flg_after2'] = df['holiday_flg'].shift(-2)
  df['holiday_name_after2'] = df['holiday_name'].shift(-2)

  # 金曜日や月曜日に祝日や大型連休が来る場合は、土日も連休扱いにする
  # 土曜日の連休判定
  # 金曜が祝日なら土日(祝日名なし)も連休
  df.loc[(df['dayname']=='Saturday')&(df['holiday_name'].isna())&(df['holiday_flg_before']==1), 'holiday_name'] = df['holiday_name_before']
  df.loc[(df['dayname']=='Sunday')&(df['holiday_name'].isna())&(df['holiday_flg_before2']==1), 'holiday_name'] = df['holiday_name_before2']
  # 月曜が祝日なら土日(祝日名なし)も連休
  df.loc[(df['dayname']=='Sunday')&(df['holiday_name'].isna())&(df['holiday_flg_after']==1), 'holiday_name'] = df['holiday_name_after']
  df.loc[(df['dayname']=='Saturday')&(df['holiday_name'].isna())&(df['holiday_flg_after2']==1), 'holiday_name'] = df['holiday_name_after2']

  return df

# ###########################################
# 関数定義：連休や大型連休を祝日マートに追加
# ###########################################
def add_holidays(df):
  '''
  df: DataFrame, 祝日マート
  '''
  # Prophet投入用に祝日名称は予測用に集約しておく
  # 土日の祝日名称は消しておく（単独では普通の週末と変わらない、連休の場合は後で上書きできるように）
  daynames_weekend = ['Saturday', 'Sunday']
  df.loc[df['dayname'].isin(daynames_weekend), 'holiday_name'] = None
  df.loc[df['dayname'].isin(daynames_weekend), 'holiday_flg'] = 0

  # 月曜や金曜に祝日がある場合は連休
  daynames_shumatsu = ['Monday', 'Friday']
  df.loc[(df['dayname'].isin(daynames_shumatsu))&(df['holiday_flg']==1), 'holiday_name'] = '連休'

  # 火～木曜日の祝日は単独祝日（大型連休該当の場合は後で上書き）
  daynames_nakabi = ['Tuesday', 'Wednesday', 'Thursday']
  df.loc[(df['dayname'].isin(daynames_nakabi))&(df['holiday_flg']==1), 'holiday_name'] = '単独祝日'

  # 大型連休の追加
  # 年末年始(12/29-1/3)
  df.loc[(df['mmdd']>='12-29')|(df['mmdd']<='01-03'), 'holiday_name'] = '年末年始'
  # GW（4/29-5/5）
  df.loc[(df['mmdd']>='04-29')&(df['mmdd']<='05-05'), 'holiday_name'] = 'GW'
  # お盆（8/13-16）
  df.loc[(df['mmdd']>='08-13')&(df['mmdd']<='08-16'), 'holiday_name'] = 'お盆休み'

  # 8/11の山の日が金～日曜の場合は、お盆とつながるのでお盆扱いにする
  df.loc[(df['mmdd']=='08-10')&(df['dayname'].isin(['Saturday'])), 'holiday_name'] = 'お盆休み'
  df.loc[(df['mmdd']=='08-11')&(df['dayname'].isin( ['Friday', 'Saturday', 'Sunday'])), 'holiday_name'] = 'お盆休み'
  df.loc[(df['mmdd']=='08-12')&(df['dayname'].isin( ['Saturday', 'Sunday', 'Monday'])), 'holiday_name'] = 'お盆休み'

  # 5/6が月曜日の場合は振替休日となるのでGW扱い
  daynames_furikae = ['Monday', 'Tuesday', 'Wednesday']
  df.loc[(df['mmdd']=='05-06')&(df['dayname'].isin(daynames_furikae)), 'holiday_name'] = 'GW'

  # 連休/大型連休が土日と連続する場合は大型連休に含める
  # 前後２日分の祝日フラグ/名称を更新
  df = add_before_after_holidays(df=df)

  # シルバーウィークのように例外的に国民の祝日ができる場合は個別対応
  # 単独祝日の前日が祝日の場合は連休扱い
  df.loc[(df['holiday_name']=='単独祝日')&(~df['holiday_name_before'].isna()), 'holiday_name'] = '連休'
  df.loc[(df['holiday_name']=='単独祝日')&(~df['holiday_name_before'].isna())&(~df['holiday_name_before2'].isna()), 'holiday_name'] = '連休'

  # holiday_nameが欠損ではない日のholiday_flgは全て１にする
  df.loc[~df['holiday_name'].isna(), 'holiday_flg'] = 1

  return df

# ###########################################
# 関数実行・出力
# ###########################################
# 連休や大型連休を祝日マートに追加
df_name = add_holidays(df=df_flg)
# 不要カラム削除
print(df_name.columns)
rm_cols = [
    'holiday_flg_before', 'holiday_name_before', 'holiday_flg_before2', 'holiday_name_before2', 'holiday_flg_after', 'holiday_name_after', 'holiday_flg_after2', 'holiday_name_after2'
]
df_out = df_name.drop(columns=rm_cols)