KPI Implementation guide

The purpose of this article is to describe how to get KPI values calculated in MPI. It will give overview of the process and also provide with references that might be useful for the implementation.

Where to start?

The best source of information is MPI ETL and cube help portal. You can find it on MPI DVD.

To find out how data will be mapped and transformed from the source MES system, you may follow the procedure below:

1. Identify KPI you are interested to see and locate the help article describing it.
   Find corresponding intelligence pack and navigate to "Measures and KPIs" section or "Folder Calculations" section. You will see measure definitions in "Measures in Measure Group" paragraph of measure groups or "Measures" in case of calcuations. There are two types of KPI you may find:
   Raw measures - that are derived by aggregating data straight from data warehouse fact table
   Calculation - that are calculated in cube using raw measures
   
   
   
   You can notice easily the difference between types of measures by checking "Expression" column in the table listing KPIs. Expression defines MDX formula in cube that is applied to perform calculation.
   
   
   To get calculation values in MPI, you need to make sure that formula can be calculated, meaninig that data for raw measures on which calculation depends on are available in the cube.
2. Find fact table that needs to be populated to get measure value.
   Once you identify raw measure, use the documentation to locate name of the fact table that is used to provide with data for measure group and so on for the measure.
   
   
   In the measure description you can find name of the table column that needs to be populated with data.
3. Find ETL that is used to transform data from source system to fact table.
   Knowing the fact table that needs to be filled with data, we can find ETL that is reponsible for transformig data from the source system and loading it to fact table. Navigate to "ETL Package" section of the help system. You will see list of ETLs available in intelligence pack. Usually name of the ETL corresponds to name of the fact table. Browse ETL list to make sure you can find the right ETL package.
   
   
   In the package description section, you will find general information what is the source table to pull data from. Expand other sections to find rules of transformation in "Source Description" section, "Source query" to find out how data is retrieved and "Column binding" to verify how columns are mapped between source and destination.
   
   

In case of dimensions, you may skip steps 1 and 2, you should be able to find corresponding ETL with mapping information in ETL documentation. Most of dimensions are shared objects, so if you can't find requested information under intelligence pack section, verify "Common Objects" section.