TM1 Models will all consist of 4 functional components. These will be Absorb, Configure, Calculate/Consolidate and Consume.
Remember:
- Each component must be kept distinct!
- Components are purpose based not technology based!
- Focus on which components add value to the business not ones that are generic to any business!
Now let’s have a look at each of the components.
Absorb
All models must absorb foundation information. Foundation information is the basic information that a model is trained upon; this might include sales performance results, conversion rates, business expenses, etc. Some key points:
- “Absorption” is generic and most likely will not be the “value add” to the business.
- Foundation information data sources may change in the future.
The Future of Big Data
With some guidance, you can craft a data platform that is right for your organization’s needs and gets the most return from your data capital.
IBM Cognos TM1 provides us with TurboIntegrator (TI) to load prepared foundation information into a model. It is recommended that TI be used to load data into support or “friend” cubes. These Friend cubes exist to provide support to the model removing any complexity of the foundation data from the complexity of the business problem being solved. Retrieving and preparing the data should be the responsibly of an ETL team and not part of the model architecture.
Configure
All models will require certain foundation information assumptions be configured (or” placed in context”). In addition, the model must also allow users to make foundation information adjustments (to the support various scenarios). Configuring assumptions and adjusting foundation information are generic operations and are not the business value-add. Consider using TM1 “Websheets” to set assumption parameters and make adjustments to foundation information. Most likely, a series of simple lookup cubes (LK) will be used to store the assumption configurations and track user adjustments.
Calculate and Consolidate
Training the model on foundation data (configured based on specific assumptions), is usually what is unique and will be the value add to the business. This should be the leverage-able component of the model. The model should use the Cognos TM1 Rules Engine as the real-time “engine” to do the calculation or consolidation.
Consume
Consuming the results of training the model on the calculated and/or consolidated foundation information should be considered a generic process and not really the “value add” of the model. The model should use industry standard Cognos TM1 virtual cubes to emphasize the trained information for specific user purposes and the organizations standard interface to access the compiled views.
Wrap Up
Always focus on the business “value add” and not the generic functions during model design and you’ll be a successful architect.
