合并具有开始和结束状态的连续两行事件

Question

我有一个这样的数据框：

Date	Atomic Timestamp	MAUI	Measurement Name	Operation Type	Measurement Type	Measurement Context	Measurement Action	Machine	Measurement	Action
2021-10-03	14:38:38.194376	7000001	PJ_DBO_KPI_380	0	NaN	NaN	StartJob	380	Job	Start
2021-10-03	14:38:38.836202	9200031	PJ_DBO_KPI_380	0	NaN	NaN	StartReportServiceCreate	380	ReportServiceCreate	Start
2021-10-03	14:38:39.058186	9200032	PJ_DBO_KPI_380	0	NaN	NaN	FinishReportServiceCreate	380	ReportServiceCreate	Finish
2021-10-03	14:38:39.089658	1000001	PJ_DBO_KPI_380	1	NaN	NaN	StartPickFromCarrier	380	PickFromCarrier	Start
2021-10-03	14:38:39.165627	3000402	PJ_DBO_KPI_380	1	NaN	NaN	StartWaferAlignLoadPatterns	380	WaferAlignLoadPatterns	Start
2021-10-03	15:53:50.330514	9200009	PJ_MPS_ARO	0	NaN	NaN	StartReportsCollection	1385	ReportsCollection	Start
2021-10-03	15:53:50.330540	9200010	PJ_MPS_ARO	0	NaN	NaN	FinishReportsCollection	1385	ReportsCollection	Finish
2021-10-03	15:53:50.331217	9200011	PJ_MPS_ARO	0	NaN	NaN	StartReplayDataCollection	1385	ReplayDataCollection	Start
2021-10-03	15:53:50.331243	9200012	PJ_MPS_ARO	0	NaN	NaN	FinishReplayDataCollection	1385	ReplayDataCollection	Finish
2021-10-03	15:53:53.363365	9200002	PJ_MPS_ARO	0	NaN	NaN	FinishReporting_155	1385	Reporting_155	Finish

我正在寻找一种方法来合并所有具有相同度量的行以及具有开始和结束的操作。如果与上一个开始活动相对应的该事件未完成，则排除所有行。

输出应该是这样的：

Start DT	End DT	MAUI.x	MAUI.y	Measurement Name	Operation Type	Measurement Type	Measurement Context	Machine	Measurement
2021-10-03 14:38:38.836202	2021-10-03 14:38:39.058186	9200031	9200032	PJ_DBO_KPI_380	0	NaN	NaN	380	ReportServiceCreate
2021-10-03 15:53:50.330514	2021-10-03 15:53:50.330540	9200009	9200010	PJ_MPS_ARO	0	NaN	NaN	1385	ReportsCollection
2021-10-03 15:53:50.331217	2021-10-03 15:53:50.331243	9200011	9200012	PJ_MPS_ARO	0	NaN	NaN	1385	ReplayDataCollection

Answer 1

关于合并，您可以简单地定义具有开始和结束事件的新数据集，然后将它们合并。假设 df 是您的数据框：

start_df = df[df.Action == "Start"]
# build the Start DT column
start_df["Start DT"] = start_df.Date.str.cat(start_df["Atomic Timestamp"])

finish_df = df[df.Action == "Finish"]
# build the End DT column
finish_df["End DT"] = finish_df.Date.str.cat(finish_df["Atomic Timestamp"])

merged = start_df.merge(finish_df, on=["Measurement"])

在那里，您将拥有一个数据框，其中仅包含已开始和已完成的测量。然后，您可以删除未使用的列并根据需要重新排序