我有一個 df 下面為： day mealtype ozMonday Snack .34Monday Snack .43Monday Dinner .31Tuesday Breakfast .10Monday Dinner .11Tuesday Dinner .09Sunday Snack .33Sunday Dinner .01Sunday Lunch .03我有下面的代碼計算每天每種膳食類型的百分比 df.groupby('day')['mealtype'].value_counts(normalize=True).to_frame('%').reset_index().round(1)我如何調整此代碼以獲得盎司列的分位數 - p50 和 p90，但也按膳食類型日和膳食類型分組？謝謝！
2 回答
一只甜甜圈
TA貢獻1836條經驗 獲得超5個贊
你可以試試這個
p50 = df.groupby(['day','mealtype']).agg(p50 = ('oz',lambda x: x.quantile(0.5))).reset_index()
p90 = df.groupby(['day','mealtype']).agg(p90 = ('oz',lambda x: x.quantile(0.9))).reset_index()
new_df=p50.merge(p90, on=['day','mealtype'])
new_df
或者
new_df = df.groupby(['day','mealtype'])['oz'].agg(p50 = (lambda x: x.quantile(0.5)),
p90 = (lambda x: x.quantile(0.9))).reset_index()
輸出
day mealtype p50 p90
0 Monday Dinner 0.210 0.290
1 Monday Snack 0.385 0.421
2 Sunday Dinner 0.010 0.010
3 Sunday Lunch 0.030 0.030
4 Sunday Snack 0.330 0.330
5 Tuesday Breakfast 0.100 0.100
6 Tuesday Dinner 0.090 0.090
ITMISS
TA貢獻1871條經驗 獲得超8個贊
咱們試試吧groupby().quantile()：
(df.groupby(['day','mealtype'])['oz']
.quantile(q=[0.5,0.9])
.unstack(level=-1)
.rename(columns=lambda x: f'p{int(x*100)}') # rename columns
)
輸出：
p50 p90
day mealtype
Monday Dinner 0.210 0.290
Snack 0.385 0.421
Sunday Dinner 0.010 0.010
Lunch 0.030 0.030
Snack 0.330 0.330
Tuesday Breakfast 0.100 0.100
Dinner 0.090 0.090
添加回答
舉報
0/150
提交
取消