python - 计算时间序列中的连续值

Question

这是我第一次在这里提问，所以我希望我能做对!

我有一个 Pandas 数据框:

df2.data
Out[66]: 
date
2016-01-02    0.0
2016-01-03    1.0
2016-01-04    1.0
2016-01-05    1.0
2016-01-06    0.0
2016-01-07    0.0
2016-01-08    1.0
2016-01-09    2.0
2016-01-10    1.0
2016-01-11    0.0
Name: data, dtype: float64

我想要以下结果:

            data  trend  trend_type
date                               
2016-01-02   0.0      0           0
2016-01-03   1.0      0           0
2016-01-04   1.0      1           1
2016-01-05   1.0      2           1
2016-01-06   0.0      0           0
2016-01-07   0.0      1           0
2016-01-08   1.0      0           0
2016-01-09   2.0      0           0
2016-01-10   1.0      0           0
2016-01-11   0.0      0           0

我的问题与 How to use pandas to find consecutive same data in time series 有点相关.

到目前为止，我设法掌握了趋势，但效率不够高(对于 750 行数据帧大约需要 8 秒)

df['grp'] = (df.close.diff(1) == 0).astype('int')
df['trend'] = 0
start_time = time.time()
for i in range(2, len(df['grp'])):
    if df.grp.iloc[i] == 1:
        df['trend'].iloc[i] = df['trend'].iloc[i-1] + 1 

Answer 1

第一步
要获得 trend，请执行 groupby + cumcount -

df['trend'] = df.data.groupby(df.data.ne(df.data.shift()).cumsum()).cumcount()
df

            data  trend
2016-01-02   0.0      0
2016-01-03   1.0      0
2016-01-04   1.0      1
2016-01-05   1.0      2
2016-01-06   0.0      0
2016-01-07   0.0      1
2016-01-08   1.0      0
2016-01-09   2.0      0
2016-01-10   1.0      0
2016-01-11   0.0      0

第 2 步
(IIUC)，要获取 trend_type，比较连续的行并分配。

df['trend_type'] = 0
m = df.data.eq(df.data.shift())
df.loc[m, 'trend_type'] = df.loc[m, 'data']

df

            data  trend  trend_type
2016-01-02   0.0      0         0.0
2016-01-03   1.0      0         0.0
2016-01-04   1.0      1         1.0
2016-01-05   1.0      2         1.0
2016-01-06   0.0      0         0.0
2016-01-07   0.0      1         0.0
2016-01-08   1.0      0         0.0
2016-01-09   2.0      0         0.0
2016-01-10   1.0      0         0.0
2016-01-11   0.0      0         0.0