[yang400]

I had the similar problem regarding test dataset.
From the description, “Thus each line should specify the start time of the fault, the end time of the fault, and the fault type. Each submission will receive a score, as described below.”, it is hard to decide the what is START/END time based on the existing rule. In the trainDataset, the sampling rate for XXX.C is not regular. We have no idea how do you define these time?

I had the similar problem regarding test dataset.
From the description, “Thus each line should specify the start time of the fault, the end time of the fault, and the fault type. Each submission will receive a score, as described below.”, it is hard to decide the what is START/END time based on the existing rule. In the trainDataset, the sampling rate for XXX.C is not regular. We have no idea how do you define these time?

Anothe rquestion is that the sampling for XXX.a and XXX.c is 15M for 24 hours 7 days. But for XXX.c there is no way to decide the period of the failure. From dataset xxx.C looks lie a random sample. Soem peroid of time in a day is no recordes.

It is better to provide a sample file to show the submision requirement.

One more question is why the same period of the time, a plant could be either code 6 or fault code. Here is the data for PLANT1 on sept 5, 2009.

9/5/2009 8:44 9/5/2009 8:59 6
9/5/2009 10:44 9/5/2009 11:14 1
9/5/2009 21:29 9/5/2009 23:14 1
9/5/2009 21:59 9/5/2009 23:14 6
9/5/2009 9:14 9/5/2009 23:44 6

The plant was recorded as 6 or 1 between 21:59 to 23:14. What does that mean?

Thanks,

Chunsheng

[Author]

Posts