06-20-2017 08:19 PM
I have an MOSA(marketing optimization stand alone) optimization scenario that uses a generalized customer table, which contains 2 different scores (columns) for each product. The columns look like this:
CUSTOMER_ID, REVENUE_SCORE1, PROP_SCORE1, REVENUE_SCORE2, PROP_SCORE2, .... etc.
These columns don't always represent the same product, and I am able to manage this with the Control Table, by mapping column names to communication codes and communication descriptions.
Now, I want to set up the same process in MOMA (marketing optimization + marketing automation). Is there an easy way to surface the column mapping to CI Studio, so that I know Cell X uses REVENUE_SCORE1 & PROP_SCORE1 and Cell Y uses REVENUE_SCORE2 & PROP_SCORE2?
I know I could maintain a mapping file locally and assign the optimization settings by hand, but I want to know if there is an easy way to set up something similar to the Control Table in MOSA, for MOMA.
Thanks in advance.
06-20-2017 10:56 PM - edited 06-20-2017 10:56 PM
In MOMA you define the measure names in the business context, and in the Optimization Group, for each measure you defined, you choose the column from your information map that represents the score to be used for that cell. The control table is dynamically generated by the software when you generate the input data for MO.
06-23-2017 11:36 AM
Thanks shill. I guess my question then becomes: is there an easy way to dynamically produce and modify the info map? So that generalized column names show up with more descriptive names in CI Studio, without having to modify data items one at a time in info map studio?
06-23-2017 04:18 PM
06-26-2017 08:40 AM - edited 06-26-2017 08:40 AM
Well, PROC INFOMAPS is an underlying tool that consultants use to generate complex information map based on a data-driven definition (often coming from Excel files). You don't use the procedure directly, there exist quite complex wrapper scripts that take care of all aspects of the infomap generation (using "proc infomaps" as a core) - source definitions, table relations, columns, custom attributes etc. I heard one such script is being used in SAS, and we in T1A have built a similar script for our consultants.
It is possible to use the same tool for a simple column addition to an existing informap, however I don't think the process can be fully automated. You'll have to go to CI Studio to refresh a business context anyway, and I don't think it is a good idea to do so while other users are in the systems and campaigns may be running. So, if you only need to add one new column, the programmatic solution using proc infomaps might not be easier than adding the same column manually through SAS Information Map Studio.
There is another way however, which may or may not be beneficial in different circumstances.
If the model scores are added to the data mart in a vertical (narrow) table: "CUSTOMER_ID", "MODEL_NAME", "SCORE_VALUE", it is possible for the users to define calculated items that extract one model value at a time:
max(case when <<MODEL_NAME>> = "MY MODEL 123" then <<SCORE_VALUE>> else NULL end)
Numeric calculated items with an aggregation function behave as measures, so get aggregated over a data set (group by subject ID), that's why it should work.
The process of defining such calculated items is still manual, but at least it can be delegated to the end-users. This may be helpful in certain circumstances (not always through).