Compensate missing value not working as expected....

[Chee Kheong Wan]

I have create a time interval property selection for hourly and 2 hourly...but the features for compensate missing value only works in hourly internal not the 2 hourly internal.

When choose 2 hourly interval, the line charts become dots...

Anyone got any solution?

thanks in advance...

[Gaia Paolini]

If you go into Properties > Appearance and untick "Break line on empty values", the lines appear.

[Chee Kheong Wan]

have try that but it does not work...

[Gaia Paolini]

you need to untick it. It worked for me with your sample dxp without changing anything else

[Chee Kheong Wan]

The actual setting i wish to have is "Break Line with empty data" and "Compensate for missing data" both tick...

If i uncheck the "Break Line with empty data", all the data will join up where time frame suppose to have gap will be unseen.

By checked "Break Line with empty data" and "Compensate for missing data", data show ok in hourly internal but not the 2 hourly interval..

Maybe the sample i prepare data not big enough. Please refer to below attachment.

[Fredrik Rosell]

Hello,

This sounds like an expected behavior. In the 2 hourly case, compensate for missing values adds the missing hours (with your two hour intervals - 8 AM, 10 AM, 12 AM... it adds the missing hours so the the data now has entries for 8 AM, 9 AM, 10 AM, 11 AM, 12 AM...).   

As documented in the manual:

Compensate for missing values

Enabling this setting will temporarily add new rows with empty values for the missing periods so that they can be used in the calculation

Break lines on empty values

Determines whether lines should be broken or remain connected when an empty value is found in the data used to create the line.

So, with every other value empty, the line chart becomes dots with break on empty enabled.

 

[Chee Kheong Wan]

Guess this is what spofire software behave in origin design...got any work around?

Attached is the related file for check around

[Gaia Paolini]

When selected, the Time 2 Hours generates an X axis that does not have the odd hour values. So these are missing data.

By setting the X axis this way, when you compensate for missing values you generate hours that have no associated values.

In this case the options of breaking the line for missing values and compensating are incompatible (if you want to see lines).

Maybe if you give more details about the goal of showing 1- and 2-hour values, we can think of a workaround.

[Chee Kheong Wan]

In our manufacturing, different area monitoring KPOV base on different needs where time interval is one of the selection base on user options.

We are monitor data base on Hourly, 2 hourly, 6 hourly and Daily.....thus we create the property for user selection on time interval.

Also, in our nature of process, when we said data for 4pm. It is represent data pool from 3pm to 4pm for hourly interval; as for 2 hourly interval it is data pool from 2pm to 4pm.

In the origin design of Spotfire, 4pm data point is represent data pool from 4pm to 5pm and for 2 hourly interval is 2pm to 4pm.

If you refer to my dxp example, you will see what I mean.

This is the reason behind.

thanks

[Gaia Paolini]

I am not familiar with your project, so these are my thoughts, given the information I have. Hope they help, but if you think you need further discussion, it might be a good idea to contact Professional Services for more bespoke help.

Assuming that you want to see some kind of aggregation of your values over one or two hours, I would do it this way:

• keep the X axis set as the pure [TIMESTAMP] without any hierarchy
• create two versions of the values of [MDPS/CM Bot]]; one averaged over the hour ([Hourly_Value]) and one averaged over the two hours ([Two_Hourly_Value]). Then use these as additional lines.
• You might want to switch these lines on/off with some document property, similarly to what you are doing now.

I am sending you the test dxp I created, but if you have an older version of Spotfire this might not open.

So the expressions I used are:

[Hourly_Value]: Avg([MDPS/CM Bot]) over ([Test Time Hourly])

([Two_Hourly_Value]: Avg([MDPS/CM Bot]) over ([Test Time 2 Hourly])

I trellised by [Testcell] and coloured by (Column Names).

As a result of the averaging, the lines are smoothed for hourly and two hourly intervals. It might not be what you want, but without some kind of aggregation of interpolation I cannot think of a way of representing data values in chunks of time.