Friday 2 September 2011

OBIEE Naming Best Practices

Below is the obiee naming best practices:


The rules for naming of items in Analytics:


a.    Be consistent – follow a format and stick with it


b.    Be descriptive – make the meaning clear in the name


c.     Be concise – don’t go overboard with descriptive names


d.    Include Units of Measure – for metrics and numerical attributes


Try to include the aggregation rule in the name of the metric to allow easy identification of what the metric is doing.


·        Counts can use the # sign as in # Cases


·        Averages or ratios can be used as in Avg. # Open Days


·        Sums may be shown as Totals as in Total Shipped Units


·        Mins/Maxes as in Max # Shipped Units


Be very precise in the language to convey the meaning of the metric and what it is measuring:


·        # Cases Opened means something different that # Open Cases.  The former refers to a number that occurred in the past that will not change as cases close, where the latter is a current snapshot of open cases that will eventually change.


For Logical tables, the name must represent the logical role


·        Account Monthly Average Sales is a different logical entity than Account Weekly Average Sales, and as such there should be two distinct logical tables with these names.  Similar examples may include: Adapted Account vs. Registered Product and Manager vs. Employee


A general rule of data modeling is to not name attributes with the name of the entity.  For example, in an Account table, don’t have columns such as


·        Account Name


·        Account Address


·        Account Phone #


This is redundant because the fields are already organized in the Account table.  Instead, just have the following fields:


·        Name


·        Address


·        Phone #


An exception can be made for attributes whose proper name happens to include the table name.  For example, Account Number is an attribute name that should not be split, as Number by itself typically does not have any meaning.


Consider adding certain units of measure to a metric to clarify what is being measure to a user.  A metric called


·        Coverage Met does not convey enough information.  Is Coverage Met measured in weeks or days or %? Try 13 week Coverage Met or # Coverage Met


For currency metrics, consider appending the unit of currency to the metric name.


·        For example, instead of Sale Amt, consider using Sale $ or $ Sale


·        Possibly consider using Sale Amt USD for multi currency environments.


 


View the original article here

No comments:

Post a Comment