パンダのグループ初日に

Question

どのように私は1日の始まりにgroupbyを始めることができますか？

df.groupby(pd.TimeGrouper(freq='3600s')).aggregate(np.mean) 

このコードの入力：

2016-12-09 22:00:00   0.78 
2016-12-09 23:30:00   0.37 
2016-12-10 00:20:00   0.24 
2016-12-10 01:30:00   0.22 
2016-12-10 02:00:00   0.19 

このコードの出力：

2016-12-09 22:00:00   0.78 
2016-12-09 23:00:00   0.37 
2016-12-10 00:00:00   0.24 
2016-12-10 01:00:00   0.22 
2016-12-10 02:00:00   0.19 

目標出力：

2016-12-09 00:00:00   Nan 
2016-12-09 01:00:00   Nan 
... 
2016-12-09 22:00:00   0.78 
2016-12-09 23:00:00   0.37 
2016-12-10 00:00:00   0.24 
2016-12-10 01:00:00   0.22 
2016-12-10 02:00:00   0.19 

Answer 1

私はその最も簡単なだけで、あなたのデータフレームのインデックスを再作成することだと思います再サンプリングして平均を見つけると。

df['date'] = pd.to_datetime(df.date) 
df1 = df.set_index('date').resample('h').mean() 

new_idx = pd.date_range(df1.index.min().date(), df1.index.max(), freq='h') 
df1.reindex(new_idx) 

        value 
2016-12-09 00:00:00 NaN 
2016-12-09 01:00:00 NaN 
2016-12-09 02:00:00 NaN 
2016-12-09 03:00:00 NaN 
2016-12-09 04:00:00 NaN 
2016-12-09 05:00:00 NaN 
2016-12-09 06:00:00 NaN 
2016-12-09 07:00:00 NaN 
2016-12-09 08:00:00 NaN 
2016-12-09 09:00:00 NaN 
2016-12-09 10:00:00 NaN 
2016-12-09 11:00:00 NaN 
2016-12-09 12:00:00 NaN 
2016-12-09 13:00:00 NaN 
2016-12-09 14:00:00 NaN 
2016-12-09 15:00:00 NaN 
2016-12-09 16:00:00 NaN 
2016-12-09 17:00:00 NaN 
2016-12-09 18:00:00 NaN 
2016-12-09 19:00:00 NaN 
2016-12-09 20:00:00 NaN 
2016-12-09 21:00:00 NaN 
2016-12-09 22:00:00 0.78 
2016-12-09 23:00:00 0.37 
2016-12-10 00:00:00 0.24 
2016-12-10 01:00:00 0.22 
2016-12-10 02:00:00 0.19