Create Counter Widget

hieuiph 2017-03-27 06:21:07 UTC #1

Hi all,

I am trying to do the dashboard : "Cumulative count of the number of anomalies as a counter"

Char type is "COUNTER"

Value fileds for Y-axes: anomaly_count.

Value filed for X-axes: timestamp

Display Field: timestamp

Filter Data: I choose "anomaly_count" with "criteria" :above 0

and choose also "Show Values on Viewer"

but when apply, the dashboard didn't appear as expect.

please advice. Thanks

anupam.jain 2017-03-27 13:43:22 UTC #2

Some how prediction sample input data is not saved in pipeline, may be inspect had run when first time its data uploaded and then saved the pipeline. I have re-uploaded the sample data and saved the pipeline. Kindly run the pipeline, create fresh widget and verify.

hieuiph 2017-03-28 09:39:27 UTC #3

Hi,

I tried again the counter of anomaly_count but get the error below:

thanks

anupam.jain 2017-03-28 13:29:46 UTC #4

Kindly change the check point path on save poppup of pipeline and then run and verify widget. Otherwise it will process form last pipeline run state where whole data processed and next data will not be available.

Currently I set the path to /rc_2810 to verify widget. You can increment the 2810 to 2811 and save and run pipeline

hieuiph 2017-03-30 09:28:41 UTC #5

Hi,

I followed your suggestion and now I can create widget for counter of anomaly as below:

I don't know what field should be chosen for X-axe, so I leave blank. The filter is anomaly_count > 0.

Is that logic ?

thanks

anupam.jain 2017-03-30 09:31:59 UTC #6

Yes it is fine... Are you able to see graph on widget?

hieuiph 2017-03-30 09:47:01 UTC #7

you mean that the line of every anomaly detected ? I am a little bit confused what filed should be chosen. I see have only anomaly_count filed suitable but it's cumulative variable

hieuiph 2017-03-30 09:48:31 UTC #8

one more question: If we save the widget, and then we run again the pipeline, can we use the saved widget or we have to create a new one ?

thanks

anupam.jain 2017-03-30 10:03:36 UTC #9

anomaly_count should be on Y-axis. X-axis can be blank or for time wise visualization you can choose timestamp in X-axis. The timestamp is current time of data processed, it is long values in milli seconds. Cumulative value will increase with time.

Yes you can use the saved widget.

hieuiph 2017-03-30 10:10:18 UTC #10

Hi,

I get the error below when creating the line chart of each anomaly:

The configuration is :

Y_axis: anomaly_count

X-axis : timestamp

filter: anomaly_count > 0

thanks

hieuiph 2017-03-30 10:34:28 UTC #11

Hi,

If I don't choose the timestap for X-axis, it works.

If choose timestap, error "empty string" occurred.

hieuiph 2017-03-31 02:08:25 UTC #12

Hi,

any advices for this issue ?

thanks

anupam.jain 2017-03-31 16:26:35 UTC #13

empty string is shown, because cumulative aggregation has only anomaly_count as output field and timestamp is blank.

hieuiph 2017-03-31 17:35:50 UTC #14

Hi,

I modified the cumulative aggregation by adding the current timestamps. and then, I tried again the widget plot (with timestamp filed now), and it shows the timestamps but always 31/03/2017, is today. so it's normal,right ?

thanks

hieuiph 2017-03-31 17:47:51 UTC #15

timestamps don't showed as minutes or seconds ? it shows as days

anupam.jain 2017-04-03 04:14:43 UTC #16

It is fine as per the contest, but if you really want, then you can create a new text non input field and enrich it with current data and time having format that includes minutes and seconds also(using java syntax).

hieuiph 2017-04-03 04:16:59 UTC #17

thanks. and this is acceptable for contest context, that's right ?

Best Regards,

anupam.jain 2017-04-03 04:18:12 UTC #18

yes it is fine

Thanks