cancel
Showing results for 
Search instead for 
Did you mean: 

Grouped Rule Scores

Former Member
0 Kudos

So here's my context....

I have created a view in Data Insight.

I have created a rule, that scores when column [X] is blank.

The Rule Score is added to my scorecard.

All this works lovely

Now,

The view also has a column called 'Category' and contains 4 values A,B,C & D.

What I would like, is to now modify my Rule/Score, so that I get a score for each category A,B,C,D.

Obviously I could create 4 Views, each one filtering on each of the values ABCD.

However, I am thinking that a new category might get added in the future.

Or, that another situation I may have more categories than I care to create views for !

So  .. is there a way to simply get InfoSteward to 'split' the score by whatever values are in my 'Category' column ?

Any advice/comments appreciated.

Of course, I'd also appreciate knowing if this just is not do-able !

If the 'ideal' is not available, I'll take whatever I can

Thanks

Simon

Accepted Solutions (0)

Answers (2)

Answers (2)

Former Member
0 Kudos

Another member asked a somewhat similar question recently on this thread: 

Basically there is no out of the box functionality for that score breakdown; you can either write different rules that filter for your different categories and bind each rule to the same table/view, or you can create different views that filter for each category and bind a single rule to multiple views.  Both methods have pros and cons. 

The crux of the issue is that Information Steward does not store the score "denominator" (record count) by category, so it would be impossible to calculate the score by category unless you manually determined that somewhere else.  In that case, you could use the Failed Data Repository information to construct your own scores outside of Information Steward.

Hope that helps!

0 Kudos

Hello Simon,

Can you describe the rule logic that you would like to implement for that column? Is it to get a count of how many number of times A repeats?

I would think that you can create different rules that only look for those categories, and then bind them to that column.

For example you could make the rule look for instances where Category equals A, and all the results would be a failed record.

BEGIN RETURN

$parameter0 != 'A';

END

And after that you would be able to add multiple rules to your scorecard.

However, for more complex logic you would need to use nested statements, but without knowing the logic that you would want to implement this might be overkill.

Former Member
0 Kudos

Vitali,

The rule is thus:

     BEGIN RETURN

     length($Plant) >= 1;

     END

In real life, the 'Category' is a Continent, so I have 4 rules, each with a unique a filter thus :

$Continent = 'Asia'

$Continent = 'Europe'

$Continent = 'North America'

$Continent = 'South America'

This works.

However, my next project involves finding the number of errors per distibution warehouse.

There are hundreds of distribution locations.

That means I will need hundreds of rules.

This simply is not reasonable to do.

I will have to build 1 rule, and then query the failed validation repository in order to group by warehouse.

However, then I'm defeating the point of using IS and scorecards ......

We really need a way to 'group' scores within IS.

former_member655569
Participant
0 Kudos

Hi Simon,

Try if 'Dependency profiling' on this table works for you. You can have 'distribution location' as primary column and error column as dependent column. Below link has details on setting up profiling.

Information Steward 4.0: Perform Dependency Profiling

- Dhareppa