如何在 Python 中创建填充事件之间时间的行

Question

我正在构建从 2018-01-01 00:00:00 开始到今天结束的生存分析数据框架。对于与 ID 相关联的事件，我有两列 only 的开始和结束时间。

但是，我需要添加时间 between 事件未观察到

我在这里展示我所拥有的：

+--------+-----+-----+---------+------ ---------------+ |状态 | ID1 | ID2 |开始时间 |结束时间 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-04 04:00:00 | 2019-12-04 19:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-08 06:30:00 | 2019-12-20 10:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-22 11:00:00 | 2019-12-22 23:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-26 08:00:00 | 2019-12-29 16:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-19 08:00:00 | 2018-09-20 04:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-25 16:30:00 | 2018-09-26 23:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-27 01:30:00 | 2018-09-27 10:30:00 | +--------+-----+-----+---------+------ ---------------+

而我需要的是：

+--------+-----+-----+---------+------ ---------------+ |状态 | ID1 | ID2 |开始时间 |结束时间 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2018-01-01 00:00:00 | 2019-12-04 04:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-04 04:00:00 | 2019-12-04 19:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-04 19:30:00 | 2019-12-08 06:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-08 06:30:00 | 2019-12-20 10:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-20 10:00:00 | 2019-12-22 11:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-22 11:00:00 | 2019-12-22 23:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-22 23:00:00 | 2019-12-26 08:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-26 08:00:00 | 2019-12-29 16:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 111 | AA1 | 2019-12-29 16:30:00 |今天 | +--------+-----+-----+---------+------ ---------------+ |状态1 | 112 | AA1 | 2018-01-01 00:00:00 | 2018-09-19 08:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-19 08:00:00 | 2018-09-20 04:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA1 | 2018-09-20 04:30:00 | 2018-09-25 16:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-25 16:30:00 | 2018-09-26 23:00:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA1 | 2018-09-26 23:00:00 | 2018-09-27 01:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-27 01:30:00 | 2018-09-27 10:30:00 | +--------+-----+-----+---------+------ ---------------+ |状态2 | 112 | AA2 | 2018-09-27 10:30:00 |今天 | +--------+-----+-----+---------+------ ---------------+

我已经尝试过这段代码（借自：How to find the start time and end time of an event in python?），但它只给了我事件的顺序，而不是所需的行和@Fredy Montaño 提供的答案（下）：

fill_date = []
for item in range(1,df.shape[0],1):
    if (df['End_Time'][item-1] - df['Start_Time'][item]) == 0:
        ""
    else: 
        fill_date.append([df["State"][item-1], df["ID1"][item-1], df["ID2"][item-1], df['End_Time'][item-1],df['Start_Time'][item]])

df_add = pd.DataFrame(fill_date)
df_add.columns = ["State", "ID1", "ID2", 'Start_Time', 'End_Time']

df_output = pd.concat([df[["State", "ID1", "ID2", "Start_Time", "End_Time"]], df_add],axis = 0)
df_output = df_output.sort_values(["State", "ID2", "Start_Time"], ascending=True)

我认为我必须对 STATE、ID1 和 ID2 变量设置一个条件，以免占用前一组的时间。

有什么建议吗？

Answer 1

也许这个解决方案适合你。 我切片数据框只是为了获取日期，但它对你有用，你可以在考虑到状态和 ID 的情况下重复它

df = df[['Start_Time', 'End_Time']]

fill_date = []
for item in range(1,df.shape[0],1):
    if df['Start_Time'][item] - df['End_Time'][item-1] == 0:
        ""
    else: 
        fill_date.append([df['End_Time'][item-1],df['Start_Time'][item]])

df_add = pd.DataFrame(fill_date) 
df_add.columns = ['Start_Time', 'End_Time']

最后，我做了一个 concat 将原始数据框与未观察到的事件日期的新 df dates on squares are the new

一起加入您的行列

df_final = pd.concat([df,df_add],axis = 0)
df_final.sort_index(0)