[英]finding first instance of condition in all the groups using groupby in pandas
我試圖找到每個組中滿足條件的第一個實例,然后將不同的組合並在一起。 在下面的數據中,我希望當“ putbuy”列中的每個月的數據從0變為1時(即1994-2018),將單獨的列中的第一個實例設為True
。
數據:
month_x year_x day_x putbuy Desired putbuy
5977 2 2018 14 1 1
5978 2 2018 15 1 0
5979 2 2018 16 1 0
5980 2 2018 19 1 0
5981 2 2018 20 1 0
5982 2 2018 21 1 0
5983 2 2018 22 0 0
5984 2 2018 23 1 0
5985 2 2018 26 0 0
5986 2 2018 27 1 0
5987 2 2018 28 0 0
5988 3 2018 1 0 0
5989 3 2018 5 0 0
5990 3 2018 6 0 0
5991 3 2018 7 0 0
5992 3 2018 8 0 0
5993 3 2018 9 0 0
5994 3 2018 12 0 0
5995 3 2018 13 0 0
5996 3 2018 14 0 0
5997 3 2018 15 0 0
5998 3 2018 16 0 0
5999 3 2018 19 1 1
6000 3 2018 20 1 0
6001 3 2018 21 0 0
6002 3 2018 22 1 0
6003 3 2018 23 1 0
6004 3 2018 26 1 0
6005 3 2018 27 0 0
6006 3 2018 28 0 0
解決方案嘗試:
grouped=options.groupby(['month_x','year_x'])
for group in grouped:
while 'Close_x'>'pstrike':
putb=0
else:
putb=1
break
print(group)
我的數據集快照:
IIUC,您可以使用idxmax
來查找“ putbuy”最大值的首次出現的索引:
df.loc[df.groupby(['year_x','month_x'])['putbuy'].idxmax(),'DO'] = 1
df['DO'] = df.DO.fillna(0).astype(int)
print(df)
輸出:
month_x year_x day_x putbuy Desired putbuy DO
5977 2 2018 14 1 1 1
5978 2 2018 15 1 0 0
5979 2 2018 16 1 0 0
5980 2 2018 19 1 0 0
5981 2 2018 20 1 0 0
5982 2 2018 21 1 0 0
5983 2 2018 22 0 0 0
5984 2 2018 23 1 0 0
5985 2 2018 26 0 0 0
5986 2 2018 27 1 0 0
5987 2 2018 28 0 0 0
5988 3 2018 1 0 0 0
5989 3 2018 5 0 0 0
5990 3 2018 6 0 0 0
5991 3 2018 7 0 0 0
5992 3 2018 8 0 0 0
5993 3 2018 9 0 0 0
5994 3 2018 12 0 0 0
5995 3 2018 13 0 0 0
5996 3 2018 14 0 0 0
5997 3 2018 15 0 0 0
5998 3 2018 16 0 0 0
5999 3 2018 19 1 1 1
6000 3 2018 20 1 0 0
6001 3 2018 21 0 0 0
6002 3 2018 22 1 0 0
6003 3 2018 23 1 0 0
6004 3 2018 26 1 0 0
6005 3 2018 27 0 0 0
6006 3 2018 28 0 0 0
# create a copy of data
tmp_df = options.copy()
# take diff from previous day
tmp_df.loc[:, 'putbuy_change'] = tmp_df.groupby(['month_x', 'year_x']).putbuy.diff(1)
# keep rows where change is 1
keep = tmp_df[tmp_df.putbuy_change == 1]
# keep first instance of each month
first_ins = keep.groupby(['month_x', 'year_x']).head(1)
# add desired boolean indicator
first_ins.loc[:, 'result_col'] = True
# merge back onto data
result_df = options.merge(first_ins[['month_x', 'year_x', 'day_x', 'result_col']], on=['month_x', 'year_x', 'day_x'], how='left')
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.