與熊貓按日期分組DatetimeIndex

Question

我已經將數據框索引為DateTimeIndex，如下所示：

timstamp                 _id
2018-05-09 16:56:40.940  somedata1
2018-05-09 16:54:03.959  somedata2
2018-05-10 16:53:42.975  somedata3
2018-05-11 16:52:44.897  somedata4
2018-05-11 16:46:35.902  somedata5

我想顯示每個日期的頻率，如下所示：

day                      count
2018-05-09               2
2018-05-10               1
2018-05-11               2

提前致謝！

Answer 1

這是一種方式。

# convert to datetime
df['timestamp'] = pd.to_datetime(df['timestamp'])

# normalize, count values, convert to dataframe
res = df['timestamp'].dt.normalize()\
                     .value_counts()\
                     .to_frame().reset_index()

# rename columns
res.columns = ['timestamp', 'count']

結果：

print(res)

   timestamp  count
0 2018-05-09      2
1 2018-05-11      2
2 2018-05-10      1

Answer 2

str.split + groupby + count

df.groupby(df['timstamp'].str.split().str[0])._id.count().reset_index()

     timstamp  _id
0  2018-05-09    2
1  2018-05-10    1
2  2018-05-11    2

---

to_datetime + groupby + count

df.assign(
    timstamp=pd.to_datetime(df['timstamp']).dt.floor('D')
).groupby('timstamp', as_index=False)._id.count()

要么，

df['timstamp'] = pd.to_datetime(df['timstamp']).dt.floor('D')
df.groupby('timstamp', as_index=False)._id.count()

    timstamp  _id
0 2018-05-09    2
1 2018-05-10    1
2 2018-05-11    2